<%BANNER%>

Intraindividual Variability in Objective and Subjective Sleep and Cognitive Performance in Older Adults with Insomnia


PAGE 1

1 INTRAINDIVIDUAL VARIABILITY IN OB JECTIVE AND SUBJECTIVE SLEEP AND COGNITIVE PERFORMANCE IN OLDER ADULTS WITH INSOMNIA By JOSEPH M. DZIERZEWSKI A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Joseph M. Dzierzewski

PAGE 3

3 To my Mom and Dad for everything.

PAGE 4

4 ACKNOWLEDGMENTS There are so many people that have contributed to this project in one way or another. First, and foremost, I must thank my chair, Dr. Chri stina McCrae, for her mentorship. She has given me the support I needed to grow as a scientist. I must also thank my co-chair, Dr. Michael Marsiske, for his encouragement and guidance. I ha ve been truly fortunate to have two mentors willing to take part in my educational experience. I also express my gratitude to the team of researchers in the Sleep Research Lab: Amanda Ross, Joseph MacNamara, Natalie Dautovich, Ashley Stripling, and countless under graduate research assistants. At this point I have to turn my appreciation to my parents, Steve and Karen Dzierzewski, for never holding back. They were always there, always caring, and al ways believing in me. They have given me everything and asked for nothing but my happiness in return. I couldnt have asked for better parents. I owe them everything. At this point I have to turn my appreciation to my parents, Steve and Karen Dzierzewski, for never holding back. They were always there, always caring, and al ways believing in me. They have given me everything and asked for nothing but my happiness in return. I couldnt have asked for better parents. I owe them everything.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES................................................................................................................ .........9 ABSTRACT....................................................................................................................... ............10 CHAPTER 1 INTRODUCTION..................................................................................................................12 2 REVIEW OF THE LITERATURE........................................................................................14 Older Adults................................................................................................................... .........14 Sleep in Older Adults.......................................................................................................... ....14 Normal Sleep...............................................................................................................14 Insomnia....................................................................................................................... ...15 Sleep and Cognition............................................................................................................ ....16 Sleep and Cognition in Older Adults...............................................................................20 Intraindividual Variability (IIV).............................................................................................21 Variability in Sleep..........................................................................................................22 Variability in Cognition...................................................................................................23 Summary........................................................................................................................ .........26 3 STATEMENT OF THE PROBLEM......................................................................................28 Aim 1: To Determine the Amount of Variab ility in Sleep and Cognition Found Within Older Insomniacs Compared to the Amount of Variability Between Older Adults with Insomnia....................................................................................................................... .......29 Importance..................................................................................................................... ..29 Hypothesis..................................................................................................................... ..29 Analysis....................................................................................................................... ....29 Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive Measures in Older Adults with Insomnia...........................................................................30 Importance..................................................................................................................... ..30 Hypothesis..................................................................................................................... ..30 Analysis....................................................................................................................... ....30 Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-toDay-Level) Association betw een Sleep and Cognition in Older Adults with Insomnia.....31 Importance..................................................................................................................... ..31 Hypothesis..................................................................................................................... ..31 Analysis....................................................................................................................... ....31

PAGE 6

6 4 METHODS........................................................................................................................ .....36 Main Study..................................................................................................................... .........36 Participants................................................................................................................... ..........36 Inclusion Criteria.............................................................................................................36 Exclusion Criteria............................................................................................................37 Sample Characteristics....................................................................................................37 Measures....................................................................................................................... ..........38 Sleep Measures................................................................................................................38 Subjective sleep measures........................................................................................38 Objective sleep measures.........................................................................................39 Cognitive Measures.........................................................................................................40 Reasoning.................................................................................................................40 Processing speed......................................................................................................41 Procedures..................................................................................................................... ..........41 Baseline Study Procedure................................................................................................41 Alternate Forms of th e Cognitive Measures....................................................................42 Missing Data................................................................................................................... .43 5 RESULTS........................................................................................................................ .......46 Aim 1: To Determine the Amount of Variab ility in Sleep and Cognition Found Within Older Adults with Insomnia Compared to the Amount of Variability Between Older Adults with Insomnia..........................................................................................................46 Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive Measures in Older Adults with Insomnia...........................................................................47 Associations of With in-Person Variability......................................................................47 Associations of Mean-Level Performance......................................................................48 Associations between Within-Person Vari ability and Mean-Level Performance...........48 Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-toDay-Level) Association betw een Sleep and Cognition in Older Adults with Insomnia.....49 Multicollinearity..............................................................................................................49 Within-person multicollinearity...............................................................................50 Between-person multicollinearity............................................................................50 Multilevel Model for Letter Series..................................................................................50 Multilevel Model for Symbol Digit.................................................................................51 6 DISCUSSION..................................................................................................................... ....64 Review of Findings............................................................................................................. ....64 Aim 1: To Determine the Amount of Va riability in Sleep and Cognition Found Within Older Adults with Insomnia Compared to the Amount of Variability Between Older Adults with Insomnia..........................................................................64 Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive Measures in Older Adults with Insomnia....................................................................64

PAGE 7

7 Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Dayto-Day-Level) Association between Sleep and Cognition in Older Adults with Insomnia.......................................................................................................................66 MLM for reasoning..................................................................................................66 MLM for processing speed......................................................................................67 Study Limitations.............................................................................................................. ......68 Theoretical and Empirical Implications..................................................................................69 Future Directions.............................................................................................................. ......72 APPENDIX A SLEEP DIARY.................................................................................................................... ...73 B LETTER SERIES TEST.........................................................................................................74 C SYMBOL DIGIT TEST.........................................................................................................76 D CORRELATIONS OF PREDICTOR VARI ABLES NOT CONTROLLING FOR TIME...78 E REASONING MODELS AT EACH STEP OF MLM BUILDING......................................79 F PROCESSING SPEED MODELS AT EA CH STEP OF MLM BUILDING........................82 LIST OF REFERENCES............................................................................................................. ..85 BIOGRAPHICAL SKETCH.........................................................................................................93

PAGE 8

8 LIST OF TABLES Table page 3-1 Level 1 and Level 2 Equations at ea ch Step of the MLM Building Process....................35 4-1 Participant Characteristics............................................................................................... .45 5-1 Amount of within and between-person variability...........................................................53 5-2 Correlations of within-p erson standard deviations...........................................................55 5-3 Correlations of sample means...........................................................................................56 5-4 Correlations of with in-person standard deviations and sample means..............................57 5-5 Correlations of with in-person predictors...........................................................................58 5-6 Correlations of between-person predictors........................................................................59 5-7 Steps taken in building the Letter Series Multilevel Model..............................................60 5-8 Sleep variables predicting reasoning.................................................................................61 5-9 Steps taken in building the Symbol Digit Multilevel Model.............................................62 5-1 Sleep variables pred icting processing speed......................................................................63 A-1 Example of a Sleep Diary................................................................................................. .73 D-1 Correlations of within-p erson, with time, predictors.........................................................78 D-2 Correlations of between-p erson, with time, predictors......................................................78 E-1 Reasoning MLM St ep 2: Adding Time.............................................................................79 E-2 Reasoning MLM Step 3: Adding Sleep.............................................................................80 E-3 Reasoning MLM Step 4: Adding Interactions...................................................................81 F-1 Processing Speed MLM Step 2: Adding Time..................................................................82 F-2 Processing Speed MLM Step 3: Adding Sleep..................................................................83 F-3 Processing Speed MLM St ep 4: Adding Interactions........................................................84

PAGE 9

9 LIST OF FIGURES Figure page 4-1 Overall design of REST Study..........................................................................................44 5-1 Relative amount of w ithin-person variability...................................................................54

PAGE 10

10 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science INTRAINDIVIDUAL VARIABILITY IN OB JECTIVE AND SUBJECTIVE SLEEP AND COGNITIVE PERFORMANCE IN OLDER ADULTS WITH INSOMNIA Joseph M. Dzierzewski May 2007 Chair: Christina McCrae Cochair: Michael Marsiske Major: Psychology Sleep researchers and theorists alike have speculated for many years that one of the primary purposes of sleep is to provide a period of rest for the brain that is essential for maintaining optimal cognitive functioning. Altho ugh many have attempted to systematically examine this relationship, the literature is pl agued by inconsistent and contradictory findings. The current study examined the sleep-cognition re lationship in a sample of community-dwelling older adults with insomnia. Forty-eight older adults with insomnia (M ean Age = 69.91 years, SD = 7.24) concurrently wore wrist actigraphy while completing sleep diar ies and a measure of perceptual speed and reasoning for fourteen consecutive days. Descriptive analysis reveal ed that individuals exhibited a substantial amount, between 48% and 189%, of va riability in sleep and cognition within-person as compared to between-person. Correlational anal ysis revealed that more variable sleep is associated with poorer reasoning and better perceptual speed. Multilevel modeling (MLM) indica ted that days with longer th an average sleep onset are associated with better than average reasoning, as is spending, on average, gr eater time resting in the morning. However, days when an individual, who usually spends long periods to fall asleep, takes even longer to fall asleep are associat ed with poorer reasoning. Spending above-average

PAGE 11

11 amount of time awake in the middle of the night is associated with below-average perceptual speed. However, days when an individual, w ho usually spends long periods resting in the morning, rests for even longer in the morning are associated with better processing speed. Results lend support to the sleep-cognition rela tionship. These findings illustrate the utility of studying intraindividual variab ility in sleep. There is a dail y relationship between sleep, or better stated wake, parameters and cognitive performance in older adults with insomnia.

PAGE 12

12 CHAPTER 1 INTRODUCTION The current study seeks to examine the intrai ndividual variability (i.e., within-person variability) of objectively and s ubjectively measured sleep and c ognitive performance in older adults with insomnia. Specifically, this study wi ll address questions concerning the amount of within-person variability in sleep and cognition, how this vari ability might be related across the two domains and how it is related to average pe rformance/behavior, and how changes in daily sleep might relate to changes in daily cognitive performance. There is a large amount of re search illustrating that norma l aging is associated with decreases in sleep and decreases in cognitive ability. Given the known decreases in sleep and cognitive performance associated with increased age, it is logical to think that these two phenomena might be related. In fact many studies have proven this to be an accurate assumption by means of sleep depriving indivi duals and then testing their c ognitive capabilities. However, the ecological validity of such studies is low because individuals do not tend to actively keep themselves awake for such extended periods of time. And, though generally decrements in functioning are observed following deprivation, some studies ha ve found no such results. An interesting interpretation of these mixed results is that people vary in their susceptibility to sleep loss (i.e., there are interindividual differe nces in intraindivi dual processes). A current line of research, and analysis, within the developmental and cognitive aging fields is the study of intraindivi dual variability in cognitive func tioning. Results from this context have indicated that within-person variability is both a normal homeostatic hum and large inconsistency can be indicative of impaired neur onal functioning and that variability tends to increase with advancing age. Furthermore, stat istical advances associated with the study of intraindividual variability allows for th e testing of between-p erson, within-person

PAGE 13

13 differences/associations, and between-person di fferences on within-per son processes. If individuals with impaired neural functioning exhibit high within-person variability and sleep loss is expected to cause decreased cognitive performa nce through its effects on the prefrontal cortex then the application of intraindividual variabil ity methodology to the stud y of the sleep-cognition relationship in older insomniacs is warranted and should prove fruitful. The current study should expand th e knowledge domain in the fiel ds of sleep research and cognitive aging while bridging two fields that are very complimen tary yet consistently distant from one another. It is hypothesized that sl eep and cognition will both display considerable amounts of within-per son variability, that variab ility will be related acr oss constructs, and that daily changes in sleep will coincide with daily changes in cognitive performance. These hypotheses have implications for our curre nt understanding of the purpose of sleep, of the effects of sleep loss, and causes of age -related cognitive decline. Possible implications could be sleep-based treatments to deal with one possible source of age associated cognitive decline, the need to assess sleep when conducting cognitive assessment, and the need to assess cognition multiple times to get an accurate indication of true level. The following chapters provide a comprehens ive overview of the relevant sleep and cognitive literature. This will be followed by th e specific aims, hypotheses, and analysis of each part of the study. Results will then be presented and discussed.

PAGE 14

14 CHAPTER 2 REVIEW OF THE LITERATURE Older Adults There are currently over 36 million adults living in the United States over the age of 65. This number is expected to exceed 64 million in the next 40 years and it has been estimated that the number of individuals between the age of 65 and 85 will grow by approximately 113% between 2000 and 2050 while the number of indi viduals over 85 years old will experience 388% growth in the same 50 years (U.S. Census Bure au, 2004). With this rapi dly increasing population it is essential for psychologists to expand our knowledge of the various factors affecting the aging process. Sleep in Older Adults Normal Sleep The patterns, durations, frequencies, and correl ates of sleep in older adults are all well studied phenomenon. In general, as an individual increases in age their sleep becomes lighter, shorter and more fragmented than when they were younger (Morgan, 20 00). It has been found that older adults experience more frequent shifts from one sleep stage to another, more frequent intrasleep arousals (Bosselli, Pa rrino, Smerieri, & Terzano, 1998) and more, and longer, periods of alpha activity during sleep (Webb, 1982) than do their younger counterparts. There is currently considerable debate about whether or not rapid-eye-movement (REM) sleep decreases with age (Bliwise, 1993). However, it is genera lly agreed upon that one of the most prominent changes in sleep architecture associated with the aging process is the steady and drastic decrease in the amount of time spent in deep, slow-wave-sleep (Stages 3 and 4) (SWS; Prinz et al., 1982). The decrease in the amount of time spent in SWS in late-life means that older adults spend much more time in the light, non-re storative sleep Stages 1 and 2. These changes

PAGE 15

15 have led to the sleep of older adults being char acterized as structurally lighter than that of younger adults. In fact, it has been found that older adults awaken more easily from sleep than do younger adults (Zepelin, Mc Donald, & Zammit, 1984). Whether or not these age related changes in sleep are part of the normal aging process or if they are the behavioral mani festations of an underlying pathol ogy is debatable. A recent metaanalysis of the age-related sleep changes confirm that increased ag e is associated with increased wake time after sleep onset and sleep onset latenc y and decreased total sleep time at medium to high levels (Floyd, Medler, Ager, & Janisse, 2000). In general, the functions of sleep have not yet been agreed upon but it is the consensus of the majority of sl eep researchers that the above described changes in sleep reflect n ormal ontogenetic change (Morgan, 2000). Insomnia Insomnia, on the other hand, is among the most prevalent disorders of late-life. The Diagnostic and Statistical Manual of Mental Disorders 4th Edition (DSM-IV) defines insomnia as the difficulty initiating or maintaining slee p, or non-restorative sleep, for at least one month that causes significant distress (American Ps ychiatric Association, 1994). The prevalence of insomnia in older adults is not agreed upon. However, it is known that the prevalence does increase with increasing age (O hayon, 1996). In an epidemiologica l community-based sample of over 5,000 older adults estimates of the prevalen ce of insomnia were 65% (Newman, Enright, Manolio, Haponik, & Wahl, 1997) indicat ing the widespread nature of the disorder. Furthermore, insomnia in older adults tends to be a chronic condition with the average span of the disorder lasting 12 years (McCrae et al., 2003). Insomnia in late-life is not a solitary event; it is accompanied by many unwanted consequences and correlates. Indi viduals with insomnia have been shown to experience quality of life hindrances equivalent to the experience of congestive heart failure patients (Katz &

PAGE 16

16 McHorney, 2002). Other studies have reporte d deficits in cognitive and psychomotor functioning, including memory, concentration, attention, reasoning, problem solving, and reaction time (Harrison & Horn e, 2000; Roth & Roehrs, 2003). Additionally, insomnia has serious negativ e effects on mental health and social functioning. Insomnia is often co-morbid with psychiatric, mood and anxiety disorders. Estimates suggest that 30% to 50% of individu als with insomnia also have an accompanying psychiatric disorder (Benca, 2001; Morgan, 1996) Insomnia has been associated with the occurrence of coronary heart di sease (Schwartz et al., 1999) and obesity (Vorona et al., 2005). The negative impact of insomnia does not st op at significant personal hardship. Insomnia has significant national economic consequences. In somnia has been associated with a significant increase in absenteeism due to health problems (t wice as likely to miss work), lost productivity at work, increased rates of health care utilization (after controlling for age, sex, and medical and psychiatric disease) and automobile accidents (ins omnia increases the risk of traffic accidents, accidents at home, and public accidents by 200% to 300%, and work-related accidents by 150%) (Benca, 2001; Hublin & Par tinen, 2002; Katz & McHorne y, 2002; Neubauer, 2004; Roth & Roehrs, 2003;). The total economic burden of insomnia has been estimated to be between a staggering $77 and $100 billion per year (Hubl in & Partinen, 2002; St oller, 1994; Walsh & Engelhardt, 1999). Sleep and Cognition The link between sleep and cognition has l ong been theorized and studied. Although this relationship seems rather intuitive there is a lack of consensus regarding the effects of sleep loss on cognitive functions. For reasons of experiment al control most studies of the sleep-cognition relationship use a sleep depriva tion paradigm. In this paradigm, participants are usually

PAGE 17

17 cognitively tested prior to depr ivation, kept awake in a laborator y for an extended period of time, and then tested again. In a recent review of the sleep deprivati on cognition li terature, Harrison and Horne (2000) summarize the extant mixed findings rega rding the cognitive consequences of laboratory induced sleep loss as being unclear from the lite rature whether tasks associated with cognitive speed, psychomotor skills, auditory and visual at tention, and short-term attention are sensitive to 1 night of sleep deprivation for any other reason than their monotony and lack of novelty. This view is shared by scientists who argue the sleep loss has no specific physio logical or behavioral manifestations but operates through a global re duction in arousal (Wilkinson, 1992). Support for this view has been drawn from studies illustrati ng difficult and important assessments, such as IQ test, are resilient to 36 hours or mo re of sleep depriva tion (Horne, 1988). There are, however, a growing number of neuropsychological and imaging studies that have illustrated the prefrontal re gion of the cerebral cortex is part icularly sensitive to sleep loss and that the executive functions (i.e., worki ng memory, attention, and processing speed) are dependent on processes that occur in this brai n region, suggesting that these functions are the most susceptible to sleep loss (Drummond et al ., 2000). Yet, many sleep researchers adhere to the stance that divergent skills, as opposed to convergent skills are the ones most affected by sleep loss (Harrison & Horne, 2000). And early re viewers of the sleep deprivation cognition relationship suggested a negative impact of sleep loss on reaction times and vigilance (Krueger, 1989). Recent evidence from more physiologically ba sed research suggests sleep provides an exclusive form of recovery for the cerebral cort ex, especially the prefr ontal cortex (Horne, 1993). The prefrontal cortex, which has been shown to be vital for optimal cognitive functioning, has

PAGE 18

18 been shown to recover from strains placed on it during wakefulness during sleep. This can be seen in its high voltage and slowest brainwaves compared to other br ain regions, while in NREM sleep (Muzur, Pace-Schott, & Hobson, 2002). Th e prefrontal cortex is thought to be the hardest working region of the brain and it is theref ore believed that it is also the region of the brain that needs the most recove ry during sleep. Hence, the prefr ontal cortex and its associated functions are believed to be the most suscepti ble to the loss of sleep (Durmer & Dinges, 2005). In fact, marked changes in activation of the pref rontal cortex have been observed following 3035 hours of sleep deprivation (Drummond, 2000), as were detriments in performance on a known and agreed upon task of executive functioni ng, the Tower of London Test (Horne, 1988). In a related, but different, approach Pilcher and Huffcutt (1996) performed a meta-analysis on the effects of sleep deprivati on on cognitive performance. The re sults of their analysis, which included a combined sample of over 1,900, pointed to the intriguing findi ng that not all sleep deprivation is created equal. Th ey found that partial sleep depr ivation, getting less than 5 hours of sleep a night, resulted in the largest decrease in cognitive pe rformance; more so than did short-term total sleep deprivation, less than 45 h ours of sleep deprivati on, and long-term total sleep deprivation, more than 45 hours of sleep loss. The effect size for partial sleep deprivation and cognitive performance was considerably large ( d = -3.01). Interestingly, the authors also reported large effect sizes for the relationship be tween partial sleep deprivation and simple and complex cognitive tasks (Pilcher & Huffcutt, 1996). Recent researches now suggest that total sl eep deprivation has a more profound effect on cognitive functioning than partial sleep de privation (Van Dongen, Maislin, Mullington, & Dinges, 2003). With regard to chronic partial slee p deprivation, which most realistically mimics real sleep, the effects on cogni tive performance have been mi xed. Repeated days of sleep

PAGE 19

19 restriction to between 3 and 6 hours of sleep per night have been experimentally shown to negatively affect cognitive speed, working memo ry, and attention (Belenky et al., 2003; Dinges et al., 1997; Drake et al., 2001; Van Dongen et al., 2003). Interesti ngly, it has been shown that the cumulative effect of 14 days of sleep re striction to 4 hours per night was cognitively (attention, working memory, cognitive throughput) equivalent to 2 nights of total sleep deprivation and 14 days of restrict ion to 6 hours per night was equiva lent to 1 night of total sleep deprivation (Durmer & Dinges, 2005). The cogn itive deficits observed in chronic sleep restriction do not accumulate in an additive fa shion (i.e., a total of 20 hours awake in sleep restriction over 4 days does not equal 20 hours st raight of sleep deprivation). This has been interpreted as adaptation in sleep restriction (Dra ke et al., 2001) where individuals learn to deal with the effects of sleep loss. There is a growing amount of evidence that points to the role of sleep for memory consolidation. REM sleep is susp ected to influence procedural learning and emotional memories (Gais & Born, 2004; Wagner, Gais, & Born, 2001) while NREM sleep has been implicated in declarative memory processes. Sleep may also f acilitate the realization of underlying rules or insight gaining (Wagner, Gais, Haider, Verleg er, & Born, 2004). Sleep has been shown to be essential for memory consolidation after lear ning has occurred (Walker & Stickgold, 2004) and more recently the need for sleep prior to the acquisition of information has been demonstrated. Subjects who experienced 35 hours of sleep depriv ation prior to exposure to new information had significantly worse recall of that informati on than did normal cont rol sleepers (Yoo, Hu, Gujar, Jolesz, & Walker, 2007). Sleep appears to be important for the conso lidation of previously learned information and for the acquisition of new information. See Hobson & Pace-Schott (2002) for a review of the neural system s associated with sleep and learning.

PAGE 20

20 Both the authors of the review (Harrison & Horne, 2000) and the meta-analysis (Pilcher & Huffcutt, 1996) point to the need for more rese arch on the sleep and cognition relationship in older adults. Sleep and Cognition in Older Adults The two studies that have examined the effects of sleep deprivation on cognitive performance in older adults f ound worse cognitive performance fo llowing sleep deprivation in older adults than in younger adults (Webb, 1985; Web & Levy, 1982). In a sample of over 2,500 older women with osteoporosis poor objective sleep was found to be associated with impaired overall cognitive functioning, measured by the MMSE, and decreased processing speed, measured by time to complete Trails B (Blackwell et al., 2006). Similarly, in a sample of nearly 2,000 older community-dwelling women, those who sleep less and had more difficulty initiating and maintaining sleep performed cognitively worse than those whose sleep could be characterized as good (Tworoger, Lee, Scher nhammer, & Grodstein, 2006). In a sample of over 6,000 older men, insomnia symptoms were found to be independent predictors of three year cognitive decline independent of demographic, behavioral and health factors (Cricco, Simonsick, & Foley, 2001). Although much research has been conducte d on the relationship between sleep and cognitive performance, the same can not be said for studying this relationship in older adults with insomnia. The few studies that have been completed on this sample indicate that there is a sleep cognition relationship in ol der adults with insomnia. For example, it has been found that older adults with insomnia have a deficiency in slow-wave-sleep that is related to processing speed but that this same slow-wave-sleep proc essing speed relationship is not present in normal sleeping older adults (Crenshaw & Edinger, 1999). In a study comparing good sleeping older adults to medicated and un-medicated older in somniacs on the relationship between objective

PAGE 21

21 and subjective sleep and cogni tive performance found that older insomniacs on and off benzodiazepines performed worse on measures of attention and con centration (Vignola, Lamoureax, Bastien, & Morin, 2000). In a follo w-up study, it was found that the direction and strength of the various sleep parameters relationship to cogn itive performance differed by group. In general, as sleep worsened in the good sleep ers, cognitive performance decreased. However, mixed findings were reported for both medicated and un-medicated insomniacs. As some sleep parameters increased so did c ognitive performance while other re lationships were unexpectedly found in the opposite direction (e .g., as subjective total wake tim e increased so did memory) (Bastein et al., 2003). Intraindividual Variability (IIV) For behavioral scientists to fully and effectively represent processes and change within individuals over time, the employment of in traindivdual variability methodology has been termed absolutely essential (Nesselroade, 2002). Intraindividual variability (IIV) is any change in performance over a short period of time. In this sense, it can be viewed as an individuals fluctuation around their mean. Intraindividual va riability can be con ceptualized as an individuals mean standard devi ation; that is, their average in consistency or consistency around their usual performance. The study of inraindividual variability, or with in-person variability, is vitally important for several reasons. First, it allows variance components to be decomposed into their respective within-person and between-person parts (Kreft, de Leeuw, & Ai ken, 1995). This is essential because only when it is know how much variabili ty is located within and between individuals can one attempt to explain that variability. Anothe r key reason to study intr aindividual variability is in the interpretation of the results. The ecologi cal fallacy refers to ap plying results obtained at the group-level to individuals a nd the atomistic fallacy is appl ying results obtained within an

PAGE 22

22 individual to groups (Tabachnik & Fidell, 2007). The study of intr aindividual variability allows researchers to comment on processes occurring both within individuals and between groups. Variability in Sleep Experts have commented on the relevance of individual variability in sleep patterns for broadening our understanding of sleep (Espie, 1991; Pallesen, Nordhus, & Kvale, 1998). Although IIV has received some atte ntion in the sleep literature, the majority of research has focused on characterizing variability in th e sleep patterns of good and poor sleepers. Specifically, researchers have demonstrated that individuals with insomnia tend to exhibit highly variable sleep patterns (Coates et al., 1981; Edinger et al., 1997; Edinger, Marsh, Mccall, Erwin, & Lininger, 1991; Frankel, C oursey, Buchbinder, & Snyder, 1976; Hauri & Wisbey, 1992; Vallieres, Ivers, Bastien, Beaulieu-Bonnea u, & Morin, 2005) while normal sleepers tend to exhibit less variable sleep pa tterns (Edinger et al., 1997; McCr ae et al., 2005; McCrae, Wilson, & Lichstein, 2003). In examining the relationship between bedtime and total sleep time (TST) and time in bed (TIB), it was found the approximately 50% of the total variability in th ese two variables (TST and TIB) were found within older healthy sleepers (Monk et al., 2006). However, these researchers did not attempt to systematically explain within and betw een-person variability separately. Recent research ex amining within-group variability in disordered sleepers (insomniacs, narcoleptics and indi viduals with treated and untreat ed obstructive sleep apnea) has shown that persons with sleep ap nea and narcolepsy have more variable daytime functioning and poorer cognitive functioning (c oncentration and attention) th an control good sleepers and insomniacs perform worse cognitively but have more consistent performance across a ten hour testing session (Schneider, Fuld a, & Schulz, 2004). In a 14 day diary study of 30 younger adults, it was found that variability in cognitive symptoms was predicted by sleep latency onset, and

PAGE 23

23 quality (Totterdell, Reynolds, Parkinson, & Briner, 1994). Once again, the prediction over the course of days was not separated into within and between-person components. Several studies of the effects of sleep loss on cognitive performance have examined sources of variance in various neurobehavio ral functions (Olofsen, Dinges, & Van Dongen, 2004; Van Dongen, Maislin, & Dinges, 2004). It was f ound that well over 50% of the variance in digit-symbol (ICC = .82) and psychomotor vigi lance (ICC = .69) was a product of stable interindividual differences in vulnerability to sleep loss (Van Dongen et al, 2004). Neither study made an attempt to systematically explai n the within-person variations observed. Variability in Cognition The extent literature examining within-pers on variability is vast and growing rapidly. There are, however, many inconsistencies with in the literature. For example, Salthouse, Nesselroade, and Berish (2006) examined the in traindividual variability in thirteen cognitive tasks in 113 individuals, over three occasions, be tween the ages of 18 and 97 years old. They found that within-person variab ility was approximately 50% the size of between-person variability estimates, that increased within-p erson variability was related to decreased performance, and that within-p erson variability estimates were not related across constructs (Salthouse, Nesselroade, & Be rish, 2006). However, in a si milar study Nesselroade and Salthouse (2004) examined the within-person variab ility in perceptual-motor performance in 204 individuals between the ages of 20 and 91 years old, also over three occasions. Findings from this study indicate that intraindi vidual variability is substan tial (~50% of interindividual variability), variability is trait-like (i.e., operates consistently within people over time), increased variability is associated with poorer cognitive functioning, and incr eased age is associated with increased within-person variability (Nesselroade & Salthouse, 2004).

PAGE 24

24 Within-person variability in reaction time has been found to be indicative of neurological dysfunction. Results of a study examining intraindiv idual variability indica te that adults with mild dementia show more day-to -day variability than either adults with arthritis or healthy adults. Furthermore, across groups, individuals who were more variable on one task were more variable on other tasks and mo re likely to performance wors e than their less variable counterparts. Strikingly, within -person variability wa s found to be the single-most important predictor of group membership (Hultsch, M acDonald, Hunter, Levy-Bencheton, & Strauss, 2000). Hultsch, MacDonald, and Dixon (2002) reported that in an sample of nearly 800 older adults increased age was associated with in creased within-person variability and poorer performance on several measures of cognitive ability (perceptual speed, working memory, episodic memory, and crystallized ability). They also report th at variability estimates were unique predictors of cognitive functioning independent of mean-level performance (Hultsch, MacDonald, & Dixon, 2002). The majority of the studies examining the association between increased age and withinperson variability are cross-sectional. In a six year longitudinal study of cognitive performance in healthy older adults that measured performance ev ery two years, several fi ndings of interest were reported. First, increasing age was associated wi th increased intraindivi dual variability in reaction time. Second, within-pers on inconsistency in reaction time at measure 1 was predictive of subsequent cognitive (reasoning, working me mory, processing speed, and episodic memory) decline. Third, variability increased within i ndividuals longitudinally. Fourth, at measurement occasions when individuals were more variable they also performed worse on the cognitive measures (MacDonald, Hultsch, & Dixon, 2003).

PAGE 25

25 In the study assessing intraind ividual in cognitive performanc e in older adults over the greatest number of measurement occasions (120), it was found that within-person variability was related across cognitive domains (i.e., people ar e consistently inconsiste nt) and that increased variability was associated with increased overall performance in reasoning, memory, and processing speed (Allaire & Marsiske, 2005). Th e authors, as well as other, propose that variability can serve as an adap tive function and does not always have to be an indication of deterioration (Li, Aggen, Nesselroa de, & Baltes, 2001; Siegler, 1994). In a further advancement of the literature and in an attempt to capture within-person variability in reaction times in a naturalistic se tting, Salthouse and Berish (2005) provided older adults with hand held computers that randomly prompted reaction time tests over several days. Results from this study indicate that within-per son variability is as la rge (and sometimes larger) than between-person variability but that mean-level performance on reaction time tasks are better predictors of cognitive performance (Salthouse & Berish, 2005). However, in the attempt to obtain more ecologically valid results, the author s lost much of their experiment control and noted that many unmeasured confounds could have been operating systematically within their data set. Several studies have used intraindividual variab ility indexes to categor ize different groups of individuals. Variability in affect and physical functioni ng (e.g., grip strength and finger taping) has been shown to differentiate betw een healthy individuals and individuals who sustained mild and severe head injuries. Individu als with head injuries were found to be more variable across domains and performed worse on average (Burton, Hultsch, Strauss, & Hunter, 2002). It has been found that within-person vari ability in cognitive performance (memory, speed, and fluency) is nearly as large, or larger than the between-person variabilit y in older adults with

PAGE 26

26 preclinical dementia but that the amount of within-person variability is much smaller in individuals without this disorder (Sliwinski, Hofer, & Hall, 2003) Results indicate variability may be an indicator of neurological compromise. Recently, researchers have attempted to capture the covariation, or coupling, of variables within individuals. A significant amount of with in-person variability has been found in memory failures (41%), and daily fluctuations in stressful events have been shown to predict both same day memory problems and next day memory probl ems in over 300 older adults over the course of eight days (Neupert, Almeida, Mroczek, & Spiro, 2006). In a study examining the relationship between daily stress and daily cognitive performa nce, it was found that on days when a stressful event occurred both younger and older adults ha d worse attention/concentration performances and were more variable in their cognitive perf ormances (Sliwinski, Smyth, Hofer, & Stawski, 2006). In a related study, it was found that within -person fluctuations in speed of processing predicted within-person change in memory (Sliwinski & Buschke, 2004). Summary The literature suggests the appl ication of studying intraindividua l variability in studies of the sleep-cognition relationship in older adults is warranted. To summarize this review several six key factors should be noted: 1. The number of older adults is growing rapi dly and older adults are faced with unique concerns that require sp ecialized investigations. 2. Sleep decreases in quality and amount as individuals increase in age. 3. The sleep-cognition relationship has been examined extensively through sleep deprivation studies. Although resu lts from such studies are not always consistent, it is generally agreed upon that sleep loss impairs cognitive function especially on tasks that tap the prefrontal cortex. 4. The study of intraindividual variability is esse ntial for behavior scientists to accurately portray change and proce sses within individuals.

PAGE 27

27 5. Insomniacs tend to have highly variable sl eep. Sleep researchers have commented that interindividual differences in susceptibility to sleep loss may be responsible for the inconclusive findings on the sl eep-cognition relationship. Yet, there has been a general lack of studies of intraindividual variability. 6. Research from the cognitive aging field has s hown that within-person variability is large; usually 50% of the amount of between-pers on variability, and age is associated with increased within-person variabil ity. Mixed results have been offered with regard to the stability of variability and the goodbad distinction of variability.

PAGE 28

28 CHAPTER 3 STATEMENT OF THE PROBLEM For quite a long time researchers and theorists al ike have speculated that one of the main purposes of sleep is to provide a restorative period for the brain that is essential to maintain optimal levels of cognitive functi oning. In fact, many researchers ha ve attempted to capture this sleep-cognition relationship (e.g., Harrison & Horn e, 2000; Pilcher & Huffcutt, 1996). Given the normal decrease in amount of time spent asl eep (Floyd, Medler, Ager, & Janisse, 2000), the increased prevalence of insomnia in latelife (Ohayon, 2002), and the normal cognitive decline associated with increased age (Craik & By rd, 1982; Hasher & Zacks, 1988; Lindenburger & Baltes, 1994; Salthouse, 1996), there has been a good deal of research on the sleep-cognition relationship in older adults (Bastien et al., 2003; Blackwell et al., 2006; Crenshaw & Edinger, 1999; Cricco et al., 2001; Tworoger et al., 2006 ). The conventional methods employed by the majority of these studies incl ude having participants sleep fo r one or several nights in a laboratory and complete a cognitive assessment th e following day. Assessed in this manner, the sleep-cognition relationship has been an elusive one to capture. Recent research, and advances in methodology an d analyses, in the cognitive aging field into what has been termed intraindividual va riability (IIV) has produ ced interesting results about within-person variation and relations hips between cognitive performance and many covariates (e.g. Nesselroade & Salthouse, 2004; Sliwinski & Buschke, 2004). The study of intraindividual variability has been catapulted in to the forefront of aging research by several findings indicating that variability indices are better predictors of performance than mean-level indices (Butler, Hokanson, & Flynn, 1994; Eizenman, Nesselroade Featherman, & Rowe, 1997). Recently, prominent sleep researchers have commen ted on the importance of variability in fully understanding sleep (Espie, 1991; Pallesen et al., 1998).

PAGE 29

29 The current study applies IIV methodology and an alyses to the study of the sleep-cognition relationship in older adults with insomnia. To this end, the pr esent study addresses the following three aims: Aim 1: To Determine the Amount of Variab ility in Sleep and Cognition Found Within Older Adults with Insomnia Compared to the Amount of Variability between Older Adults with Insomnia Importance Typical research focuses only on the differences found betwee n people, ignoring the fact that people themselves are often highly incons istent. If it is found that a significant amount of variability is found within-pers ons, the further examination of intraindividual variability is warranted. Hypothesis Based on prior research it is expected that th e variability within-persons will be at least 50% of that found between-persons (Nesselroade, & Salthouse, 2006). Analysis To control for any practice effects, or systema tic growth in the data, all variables (sleep and cognitive) will be de-trended prior to calc ulation of indexes of w ithin and between-person variability. To de-trend the data, linear regres sions will be ran with all sleep and cognitive variables as the dependent vari ables and time (linear, quadratic and cubic functions) as the independent variables. The subsequent unstand ardized residual values resulting from the regressions were then saved and used as time independent values. Using the residual values computed above, an index of between-person vari ability (Sample Standard Deviation, SD) and within-person variability (Individual Standard De viation, ISD) will be computed. These values will then be compared by dividing the ISD by th e SD to get the proportion of between-person variability that is found within-persons.

PAGE 30

30 Aim 2: To Determine How Variability is As sociated Across Sleep and Cognitive Measures in Older Adults with Insomnia Importance If within-person variability operates systemically , then the tendency to be variable in one domain should be associated with variability in other domains. In other words, are people generally stable or generally inconsistent? If increased within-person variability is related to increased or decreased performance then variabi lity can be labeled as either flexibility or vulnerability. Hypothesis Several studies (e.g., Allaire & Marsiske, 2002) have shown at least modest correlations among the within-person va riability estimates acro ss persons and have sugge sted that variability is related to increased performance. How sleep va riability relates to cogni tive variability in aged insomniacs and whether this variability is good is unknown. However, it is believed that variability will be related across constructs and that variability will be related to increased performance. Analysis To determine how within-person variability is related across constructs, bivariate correlations among intraindividual va riability estimates (residualiz ed ISDs to control for time effects) across measures will be run. To determin e how mean-level performance is related across constructs, bivariate correlations among mean-l evel values will be run. To determine how within-person variability relates to average performance, bivariate correlations between intraindividual variability estimates (residualized ISDs to control for time effects) and mean values will be run. An ISD analysis produces a trait-like inconsistency score for each

PAGE 31

31 individual. Such analysis addre sses questions regarding who is mo re or less consistent and what the antecedents and consequences of any between-p erson differences in inconsistency might be. Aim 3: To Determine the Between-Person (M ean-Level) and Within-Person (Day-to-DayLevel) Association between Sleep and Cogn ition in Older Adults with Insomnia Importance All prior research has examined the betw een-person association between sleep and cognition. There are no published data regardin g the day-to-day, within-person, associations between sleep and cognition, especia lly in older adults with insomnia. Howeve r, if day-to-day associations between sleep and cognition are f ound, the importance of a good nights sleep for optimal cognitive functioning would be supported. Hypothesis In general it is expected that on average better sleep will relate to better cognitive performance (between-persons, Leve l 2 Fixed Effect) and days with better sleep will relate to days of better cognitive perfor mance (within-person, Level 1 Fixed Effect). Specifically, spending long amounts of time falli ng asleep and awake in the mi ddle of the night (on average and daily) will relate to poorer executive func tioning and poorer processing speeds. Hypothesis about the relationship between terminal wakefuln ess and cognitive performance can not be made due to the lack of prior research on this sleep variable. Analysis The current aim is to examine the predictive power of within-person and between-person sleep variables on cognition performance. To acco mplish this, daily data from the objective and subjective sleep measures (SOL, WASO, TWAK) will be used to predict cognitive functioning (processing speed and executive functioning) applying a multilevel model (MLM) approach. MLM, also referred to as mixed effects modeli ng or hierarchical lin ear modeling (HLM; Bryk &

PAGE 32

32 Raudenbush, 1992), is an extension of the gene ral linear model, and does not require observations to be independent. Thus, MLMs are ve ry flexible and especially suited for daily data because of their autoregres sive nature and hierarchical st ructure with daily observations nested within each participan t (Singer, Davidson, Graham, & Davidson, 1998; Singer, Fuller, Keiley, & Wolf, 1998; Singer & Willett, 2003). Because of the hierarchical nature of our data (14 consecutive days nested within 48 participants) and in order to increase the precisi on of predicting fluctuations in processing speed and executive functioning with changes in sleep patterns, we will model the data with a MLM approach. This provides the opportunity to ex amine how well sleep predicts cognition both within(level 1) and between(l evel 2) persons. Level 1 analysis addresses questions such as: "On days in which a person reports above-avera ge sleep onset latencies, does s/he also experience lower levels of processi ng speed?" This level of analys is is concerned with questions of atypical days within an indi vidual and what predicts, within -persons, the consequences of these atypical days. Level 2 analyses examines questions like: "Do pe ople who are generally better sleepers report higher le vels of processing speed?. Both objective and subjective sleep measur es will be used to predict cognitive functioning (processing speed and executive functioning) using a fi ve-step MLM approach. Step 1 (Table 3.1, Row 1), the null (baseline) model, will estimate only a fixed and random intercept for cognitive functioning (Bryk & Raudenbush, 1992). This model will specify that cognition for person j on day i is a function of the overall group-average cognition ( 00), a between-person random error term ( u0j), and a within-person ra ndom residual component ( eij). This step provides a comparison for later models.

PAGE 33

33 In step 2, time functions ( linear) will be added as a c ovariate to the null model (Table 3.1, Row 2), producing a latent growth curve model. As such, the model will specify that cognition for person j on day i is a function of: average cognitive level ( 00), linear time (1j), a betweenperson random error term ( u0j), and a within-person ra ndom residual component ( eij). This step controls for any within-person infl ations that may be caused by a sy stematic growth in the data. Next, measures of subjective and objective sleep will be added to the model. In step 3 (Table 3.1, Row 3) the estimates of the fixed and random intercepts and fixed linear slopes for each sleep variable will be added. Thus, the daily cognition scores ( Cognitionij) for each person will be predicted by: average level of cognition ( 00), linear time (1j), the between-person effects of mean-level subjective and objec tive sleep scores, the within-p erson effects of daily-centered subjective and objective sleep scores, a between-person random error term ( u0j), and a withinperson random residual component ( eij). In step 4 (Table 3.1, Row 4), level 1 leve l 2 interaction terms of like variables (e.g., level 1 SOLS level 2 SOLS) will be entered to estimate the effect of within-person fluctuations in sleep for individuals who on averag e sleep more-or-less than others. In step 5 (Table 3.1, Row 5), the random linear slopes ( uj) of the significant daily-centered subjective and objective sleep variables and inte ractions will added in order to estimate any between-person differences in the prediction of cognition. All models will be estimated using the Maximum Likelihood (ML) method. The ability of a model to predict cognition better than the baseline model (i.e., Deviance) will be used as an index of Goodness of Fit. Improvements in predic tability will be determined by the amount of reduction of withinand between-p erson residual variances comp ared to the baseline model (Bryk & Raudenbush, 1992). Decreases in residu al and intercept va riances represent a

PAGE 34

34 proportional reduction of the predicti on error, which is analogous to R2, and will be used as an estimates of within and between-person effect sizes. The amount of agreement between model predicted values and actual values will be calculated as an estimate of an overall effect size. Because this study assessed sl eep repeatedly, and used two different methods (sleep diaries and actigraphy), issues of multicollinear ity will be a concern. Formal multicollinearity diagnostic procedures are not available for multilevel modeling. To assess for possible collinearity, a multivariate mixed-effects null mo del will be estimated. This procedure will produce both G (between-persons) and R (withi n-persons) covariance matrices, which will be subsequently rescaled into correlations using the following equation (Equation 3-1): (3-1) Correlationx,y = covariance termx,y / xvar iance yiance var.

PAGE 35

35 Table 3-1. Level 1 and Level 2 Equations at each Step of the MLM Building Process Step Level 1 equation Level 2 equation 1 Cognition ij = 0j + eij 0j = + u0j 2 Cognition ij = 0j + 1jTime + eij 0j = + u0j 3 Cognition ij = 0j + 1jTime + 2j( SOLOij SOLoj) + 3j( SOLSij SOLsj) + 4j( WASOoij WASOoj) + 5j( WASOsij SWASOj) + 6j( TWAKoij TWAKoj) + 7j( TWAKsij STWAKj)+ eij 0j = + SOLoj+ SOLsj + WASOoj + + SWASOj+ TWAKoj + STWAKj+ u0j 4 Cognition ij = 0j + 1jTime + 2j( SOLOij SOLoj) + 3j( SOLSij SOLsj) + 4j( WASOoij WASOoj) + 5j( WASOsij SWASOj) + 6j( TWAKoij TWAKoj) + 7j( TWAKsij STWAKj)+ 8j[(SOLO ij SOLoj) SOLoj] + 9j[(SOLS ij SOLsj) SOLsj] + 10j[( WASOoij WASOoj) WASOoj] + 11j[( WASOsij SWASOj) SWASO j] + 12j[( TWAKoij TWAKoj) TWAKoj] + 13j[( TWAKsij STWAKj) STWAKj] + eij 0j = + SOLoj+ SOLsj + WASOoj + + SWASOj+ TWAKoj + STWAKj+ u0j 5 Cognition ij = 0j + 1jTime + 2j( SOLOij SOLoj) + 3j( SOLSij SOLsj) + 4j( WASOoij WASOoj) + 5j( WASOsij SWASOj) + 6j( TWAKoij TWAKoj) + 7j( TWAKsij STWAKj)+ 8j[(SOLO ij SOLoj) SOLoj] + 9j[( SOLSij SOLsj) SOLsj] + 10j[( WASOoij WASOoj) WASOoj] + 11j[(WASOsij SWASOj) SWASO j] + 12j[( TWAKoij TWAKoj) TWAKoj] + 13j[( TWAKsij STWAKj) STWAKj] + eij 0j = + SOLoj+ SOLsj + WASOoj + + SWASOj+ TWAKoj + STWAKj+ u0j 1j= 1 + u1j 2j= + u2j 3j= + u3j 4j= + u4j 5j= + u5j 6j= + u6j 7j= + u7j 8j= 8 + u8j 9j= + u9j 10j= + u10j 11j= + u11j 12j= + u12j 13j= + u13j Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy. SOL = sleep onset latency, WASO = wake after sleep onset, TWAK = terminal wakefulness.

PAGE 36

36 CHAPTER 4 METHODS Main Study Data presented in this document were obtai ned during a two-week baseline assessment of sleep and cognitive performance in the initial ph ase of a randomized, controlled trial of the cognitive effects of a cognitive-behavioral interv ention for older adults with insomnia (Research and Education on Sleep and Thinking study, REST; McCrae, PI; AG024459-01). Older adults with insomnia were recruited from North centr al Florida by means of newspaper, radio, and television advertisements. At the time of anal ysis forty-eight indivi duals completed baseline assessment and were included in the present study. As stated previously, REST is a randomi zed, controlled trial. The study included two weeks of baseline (followed by randomization), four weeks of either waitlist control or insomnia treatment, two weeks of post-testing, and a two week follow-up at three months. Through each phase of the study participants completed daily sleep and cognitive measures. The overall design of the REST Study is outlined in figure 4.1. Participants Participants were recruited for participation in a clinical trial of the possible cognitive benefits of a cognitive-behavioral treatment fo r late-life insomnia. Po tential participants responded to advertisements offering a free non-dr ug treatment for insomnia for individuals sixty years of age and older. To ensure their suitab ility for the study, participants went through a thorough screening process that included ma ny inclusion and exclusion criteria. Inclusion Criteria To be included in the study pot ential participants had to be sixty years old or older and meet the diagnostic requirements of primary insomnia. These diagnostic requirements include:

PAGE 37

37 (1) self-report sleep onset latenc y or wake after sleep onset great er than thirty minutes a night, (2) sleep disturbance present at least three night s a week for six months or greater, and (3) a daytime dysfunction due to the insomnia (social, occupational, mood, or cognitive) (American Academy of Sleep Medicine, 2005; American Ps ychiatric Association, 1994). Furthermore, participants had to be either st abilized on any sleep promoting s ubstance for at least six months or not have taken any such subs tance for the past month. Exclusion Criteria Participants were excluded from the study if they reported any si gnificant medical (e.g., stroke), neurological (e.g., Pa rkinsons disease), psychological (e.g., schizophrenia, bipolar disorder) or sleep disorder other than insomn ia (e.g., sleep apnea, periodic limb movements). Participants were also excluded if they were found to have severe depression as indicated by a score of twenty-four or higher on the Beck Depression Inventory 2nd Edition (BDI-II; Beck, Steer, & Garbin, 1996) or a score of thirteen or higher on the Geriatric Depression Scale (GDS; Yesavage, 1983). To screen for potential Alzheime rs disease/dementia a cutoff score of lower than twenty-three, for individuals with a 9th grade education and higher, and a score of eighteen and lower, for individuals with less than a 9th grade education, was implemented (MMSE; Folstein, Folstein, & McHugh, 1975). The use of psychotropic or other medication known to affect sleep, such as beta-blockers, also exclude d individuals from partic ipating in the study. Sample Characteristics The sub-sample available at the time of the an alysis included fortyeight older insomniacs. The sample can be categorized as young-old as the mean age at the time of the study was 69.91 years (SD = 7.24). The sample was highly e ducated (M = 16.17 years, SD = 2.98 years), mostly married (54.3% married) and female ( 62.5 % female), and Caucasian (95.7% first language English). The sample appeared to be rela tively healthy, reporting to take an average of

PAGE 38

38 3.41 prescription medications (SD = 2.50) and 2.17 over-the-counter medications (SD = 1.92). The sample reported to be chronic insomniacs, with an average of 13.96 years of insomnia symptoms (SD = 14.54). Furthermore, participan ts reported spending an average of 133.67 minutes awake per night (SD = 100.89) and only sleeping 360.57 minutes a night (SD = 126.17). See Table 4-1 for a complete list of th e samples sleep characteristics. Measures Sleep Measures All sleep variables were measured concurren tly with a subjective measure, daily sleep diaries (Lichstein, Riedel, & Means, 1999), a nd an objective measure, wrist-worn actigraphy (Mini Mitter Co., 2001), for fourt een consecutive days during th e baseline of the REST Study. Because all of the sleep variab les were measured in two distinct ways (sleep diaries and actigraphy), an s subscript indi cates the variable was measur ed subjectively, and an o subscript indicates it was measured objectively. Subjective sleep measures Participants completed sleep diaries (Lichs tein et al., 1999) each morning for 14 days, providing subjective estimates of the following sleepwake parameters: (1) sleep onset latency (SOLs)-estimated by participants as the time it took them to fall asleep after laying down; (2) wake time after sleep onset (WASOs)estimated by participants as the total amount of time spent awake during the night; and (3) terminal wakefulness (TWAKs), computed by subtracting final wake-up time (actual time the participant awoke in the morning) from out of bed time (time participant actually got out of bed). For refe rence purposes, a sleep di ary is reproduced in Appendix A.

PAGE 39

39 Objective sleep measures Participants wore an actigraph, the Actiwatch-L, which has an integral ambient light sensor (Mini Mitter Co., 2001), on their non dominant wrist for 14 consecutive days, concurrent with the sleep diary period. The Actiwatch-L monitors ambient light exposure a nd gross motor activity and contains an omni-directional, piezoele ctric accelerometer with a sensitivity of 0.01 g-force and a light sensor with a recording range of 0.1 to 150,000 Lux. The sensors of the Actiwatch-L are sampled 32 times /second and record the peak value for each second. These peak values are then summed into 30-second activity counts. These activity counts are then downloaded to a PC a nd analyzed using Actiware-Sleep v. 3.3 (Mini Mitter Co., 2001), which uses a va lidated algorithm to identify each epoch as either sleep or wake. The software provides three default sensi tivity settings (high, medium, low). This study utilized medium sensitivity. On medium sensitivity, the threshold is set at 40 activity counts. If the total activity for an epoch was 40, it was scored as wake. If the total activity was 40, the final activity count for the epoch was based on th e level of activity in the surrounding 2 minutes (see Equation 4-1). (4-1) Total Activity Epoch A = EA-4 (.04) + EA-3 (.04) + EA-2 (.20) + EA-1 (.20) + E (2) + EA+1 (.20) + EA+2 (.20) + EA+3 (.04) + EA+4 (.04), Where A = activity count for the epoch being scor ed and EA +/1-4 = activity count in adjacent epochs. If Epoch A Total Activity (i.e., weighted sum of activity counts) exceeded the threshold of 40, then Epoch A was scored as wake ; otherwise, it was scored as sleep. Bedtime and time out of bed in the morni ng were based on sleep diary entries as recommended in the software manual (Mini Mitt er Co., 2001). Actiware-Sleep determines sleep start automatically by searching for the first 10 minutes during which no more than one epoch scored as wake. Likewise, sleep end was the last 10 minutes during wh ich no more than one epoch scored as wake. As previously menti oned, Actiware-Sleep provides objective estimates

PAGE 40

40 for all of the variables also provided by sleep diar ies. Those variables and their definitions when measured objectively by actigraphy ar e: (1) SOLo-amount of time from lay down to first sleep epochs; (2) WASOo-sum of all wa ke epochs after first sleep e poch; and (3) TWAKo-amount of time from last sleep epoch to get out of bed time. Cognitive Measures Cognition was measured in two broad domain s: processing speed and reasoning. Both cognitive domains were assessed daily via pape r and pencil, self-administered tasks for the fourteen consecutive days of baseline during the REST Study. Reasoning Inductive reasoning reflects the ability to infe r general principles fr om specific instances and apply these general principles to new instances of the proble m. Inductive reasoning is highly related to working memory (Salthouse, 1991) an d to components of executive functioning (Lezak, 1995). The letter series task (Thurstone, 1962) wa s used to measure inductive reasoning. In this task, participants have to identify the pattern fo r a series of letters. Participants are asked to choose the letter that would conti nue the established pattern (A B D A B D A B ___?) in a series of letters from five answer choices. Participan ts are given four minutes to complete as many items as possible. The maximum score is 30. Th e performance score is the number of correct responses. Over the course of the study fourteen alternate forms were given, one per day. These alternate versions of the letter series test have been shown to have high test-retest reliabilities across four blocks, each contai ning fifteen days of assessment (Allaire & Marsiske, 2005). See Appendix B for an example of the letter series test.

PAGE 41

41 Processing speed The Symbol Digit Modalities Test (Smith, 1982), which measures perceptual speed and processing speed was used. In this test, the particip ant is presented with a series of nine symbols that are paired with a single, unique digit in a key at the top of an 8.5 x 11 inch sheet. The rest of the page displays a randomized sequence of symbols with blank spots below each. The participant is instructed to write the digit that corresponds to each symbol in the key below that symbol as quickly as possible. Th e participant is given 90 seconds to place as many correct digits below their corresponding symbol. The outcome of interest is the tota l number of correct responses recorded in th e allotted 90 seconds. The SDMT has many applications and has b een administered to individuals of many different cultures, age and edu cation levels (Lezak, 1995). It has been administered both individually and in groups. The SDMT has been us ed in the detection of dyslexia, aphasia and/or cerebral dysfunction and for early screening of nor mal elders for manual motor defects, visual acuity, oculomotor coordination, and visuo-spat ial orientation difficulties (Smith, 1991). Testretest reliability, alternate form reliability, and convergent validity for the SDMT are all at or above .80 (Lezak, 2004). Over the course of the study fourteen alternate forms of the SDMT were used, one per day. These fourteen alternate versions of the symbol digit test have been shown to have high test-retest reliabilities ac ross thirty days of assessment (McCoy, 2004). See Appendix C for a reproduction of the SDMT. Procedures Baseline Study Procedure Study procedures from the REST Study baseline period that are rele vant to the present study include: an initial telephone interview, an overnight portabl e polysomnography (PSG; Medcare Diagnostics) assessment, and two in-l aboratory visits wher e actigraphic sleep

PAGE 42

42 measurement, subjective sleep measurement w ith sleep diary, and daily cognitive workbooks were collected. An initial telephone interview was conducted wi th all interested pot ential participants. During this half-hour phone in terview, demographic informa tion was collected, insomnia symptoms were assessed, medical history was taken, and an at-home visit was scheduled. At the in-home visit participants were inst ructed in the operation of the ambulatory PSG. The PSG measures blood-oxygen saturation and respir ation during sleep and was used to rule out sleep apnea as the cause of the pa rticipants sleep complaints. All pa rticipants were instructed to wear the PSG throughout the night and bring the un it in to the laboratory the following day for reading. An apnea-hypoxia i ndex (AHI) of 13.1 and higher was used as the cutoff for participation in the study. During the first laboratory visit participants were instructed in th e use of the actiwatch, sleep diaries, and daily cognitiv e workbooks (which contained the symbol digit test and letter series test). Participants were instructed to we ar the actiwatch twenty-f our hours a day and to fill out the sleep dairy daily upon awakening. A sample cognitive workbook was explained and participants were told to fill out one daily in the morning. A week from this initial meeting participants returned at which time their actiw atch data was downloaded and their sleep diary and seven cognitive workbooks were collected. The participants were then given new material and scheduled for an appointment to come back in a week and turn in the material again. Alternate Forms of the Cognitive Measures During the fourteen days of baseline assessm ent participants completed both the symbol digit tests and letter series test daily. In an attempt to minimize practice effects commonly found in repeated cognitive assessments, fourteen altern ate forms of each test were used. The alternate

PAGE 43

43 forms were constructed to be comparable in difficulty and cognitive resources needed to complete them. The alternate forms of the symbol digit tests were constructed by changing the pairings of symbols to digits in each form such that no tw o forms of the test cons isted of a digit-symbol pairing that appeared in any ot her form. This manipulation was done to prevent participants from memorizing the digitsymbol pairing. The alternate forms of the letter series test were constructed by changi ng the start letter of all letter patterns but mainta ining the underlying pattern concep t (e.g., A A B A A C A A D A A would become D D E D D F D D G D D). Pattern length and number of distracters in the answer choices was maintained across all forms. No two fo rms of the test containe d identical questions. Missing Data Throughout the two week baseline measurement period some participants forgot to wear their actiwatch for a day or f ill out their sleep diary on a given day or do one or both of the cognitive tasks. Because analyses was done using multilevel modeling (MLM), also know as hierarchical linear modeling and mixed effects modeling, missing data is not an issue. MLM allows all available data to be included in analyses, as it assumes random missing data (Bryk & Raudenbush, 1992) and therefore does not exclude whole cases (known as case-wise exclusion) due to one missing data point.

PAGE 44

44 Figure 4-1. Overall design of REST Study. Boxe d in portion, baseline, is where all study material was obtained. Baseline Randomization Treatment Control Post Test Follow-up Two Weeks Four Weeks Two Weeks Two Weeks

PAGE 45

45 Table 4-1. Participant Characteristics N Minimum Maximum Mean Std. Deviation Actigraphy Sleep Onset Latency 48 0.00 351.00 20.60 32.74 Wake After Sleep Onset 48 0.00 154.00 31.92 20.99 Terminal Wakefulness 48 0.50 211.00 18.07 25.79 Total Wake Time 48 0.00 154.00 31.94 21.01 Total Sleep Time 48 118.50 659.50 424.54 82.52 Sleep Efficiency 48 35.90 99.92 85.80 9.15 Sleep Diary Sleep Onset Latency 48 0.00 345.00 39.04 46.29 Wake After Sleep Onset 48 0.00 550.00 58.08 60.70 Terminal Wakefulness 48 0.00 270.00 31.43 42.43 Total Wake Time 48 7.00 660.00 127.80 100.23 Total Sleep Time 48 0.00 700.00 360.95 117.86 Sleep Efficiency 48 0.00 98.50 73.59 20.55 Note: All values measured in minutes.

PAGE 46

46 CHAPTER 5 RESULTS The overarching aim of the study was to ex amine the intraindividual variability in objective and subjective sleep and cognitive perfor mance in older adults with insomnia. The results are grouped and presented by the three main objectives of the study: 1. How much variability in sleep and cogni tion in older adults with insomnia is found within-persons and how much is found between-persons? 2. How is variability related acro ss the various sleep and cognition domains? 3. What are the betweenperson and within-person asso ciations between sleep and cognition in older adults with insomnia? Aim 1: To Determine the Amount of Variab ility in Sleep and Cognition Found Within Older Adults with Insomnia Compared to the Amount of Variability between Older Adults with Insomnia The first goal of the study wa s to demonstrate that there is a considerable amount of within-person variab ility in both sleep and cognition in olde r adults with inso mnia. Dissection of the sleep and cognition variables did reveal a considerable amount of variability found withinpersons. In fact, all but one variable, letter se ries, was found to displa y at least 50% of the amount of between-person variabil ity within-persons (Although the letter se ries test did not exhibit 50% of the amount of between-person variability, it did display 48% of the amount of between-person variability within-persons). Al l other variables did display at least the hypothesized 50% of between-person variability with in-persons. The number correct on the symbol digit test was found to vary within-persons 92% as much as it varies between-persons. Subjectiv ely measured sleep was also found to vary considerably within-persons. SOLS displayed 85% of the amount of between-person variability within-participants; WASOS displayed more within-person variation (113%) than betweenperson; and, TWAKS exhibited a quarter more fluctuati on within-person ( 125%) than between-

PAGE 47

47 person. All objectively measured sleep variables displayed more variability from day-to-day, within-persons, than between-persons. Variability within-persons in SOLO was found to be nearly a third larger (132%) than between-persons; WASOO displayed 112% of the amount of between-person variability within-persons; and, TWAKO exhibited nearly double the betweenperson variability within-persons (189%). For a co mpleted listing of the amount of within-person variability compared to between-person variabilit y, see Table 5-1. For a graphical depiction of the relative amount of within-per son variability, see Figure 5-1. Aim 2: To Determine How Variability is Asso ciated Across Sleep and Cognitive Measures in Older Adults with Insomnia The second objective of the study was to determin e how variability is associated across the various sleep and cognition domains. Results for th is aim will be presented in three parts. The first part will consider associations between within-person variability indices across sleep and cognition. The second part will examine the mean-l evel associations acro ss sleep and cognition. And, the third part will examine the associations between within-person and mean-level indices of sleep and cognition. Associations of Within-Person Variability Correlations of the within-person standard devi ations (residualized ISDs to control for time effects) revealed relationships between within-person variability in sleep and cognition that ranged from 0.04 to 0.40. Increased within-person variability in SOLS was significantly and positively associated with increased variabil ity in the symbol digit test (r = 0.40, p < .01) and increased within-person variability in TWAKS was significantly and positively associated with increased within-person variability in the letter series test (r = 0.14, p < .01) and symbol digit test (r = 0.35, p < .01). Increased within-person variability in SOLO was significantly and positively related to increased within-person variabili ty in the symbol digit test (r = 0.36, p < .01). Within-

PAGE 48

48 person variability in WASOO and TWAKO were both significantly and positively associated with within-person variability in the letter series te st (r = 0.26, p < .01; r = 0.26, p < .01) and the symbol digit test (r = 0.31, p < .01; r = 0.37, p < .01). Within-person variability in all sleep variables were significantly and positively associat ed with within-person variability in all other sleep variables (rs = 0.08 0.62, all p s < .05). These results suggest that inconsistency operates in a trait-like fashion within individuals a nd across sleep and cogniti on. See Table 5-2 for a complete listing of within-person variability correlations. Associations of Mean-Level Performance Correlations of mean-level sleep and cognition revealed associations that ranged from -0.06 to 0.27. Significant negative associations between mean-level sleep and number correct on the letter series test were found for SOLS (r = -0.14, p < .01), TWAKS (r = -0.12, p < .01), SOLO (r = -0.09, p < .05), and WASOO (r = -0.16, p < .01). A significant positi ve association was found between mean-level TWAKO and letter series (r = 0.12, p < .01). Significant positive associations between mean-level sleep and number correct on th e symbol digit test were found for SOLS (r = 0.27, p < .01), TWAKS (r = 0.22, p < .01), SOLO (r = 0.15, p < .01), WASOO (r = 0.12, p < .01), and TWAKO (r = 0.18, p < .01). Associations among the mean-level indices in sleep ranged from 0.04 to 0.61. Results indicate that increased unwanted wake time is associated with increased processing speed and decreased re asoning. See Table 5-3 for a complete listing of mean-level correlation coefficients. Associations between Within-Person Va riability and Mean-Level Performance Correlations of mean-level sl eep and within-perso n variability in c ognition ranged from 0.03 to 0.45. Significant positive associations between mean-level sleep and within-person variability on the number correct on the letter series test were found for SOLS (r = 0.24, p < .01), WASOS (r = 0.09, p < .05), TWAKS (r = 0.30, p < .01), WASOO (r = 0.18, p < .01), and TWAKO

PAGE 49

49 (r = 0.21, p < .01). Significant positive associations be tween mean-level sleep and within-person variability on the number correct on the symbol digit test were found for SOLS (r = 0.39, p < .01), TWAKS (r = 0.35, p < .01), SOLO (r = 0.38, p < .01), WASOO (r = 0.24, p < .01), and TWAKO (r = 0.45, p < .01). These results indicate that hi gher levels of wake time are associated with more variable cognitive functioning. Significant negative associations between within -person variability in sleep and mean-level performance on the letter seri es test were found for SOLS (r = -0.13, p < .01), TWAKS (r = -0.25, p < .01), and SOLO (r = -0.10, p < .05) indicating that inconsiste ncy in sleep is negatively related to level of reasoning. Significant positive associat ions between within-pers on variability in sleep and mean-level performance on the symbol digit test were found for SOLS (r = 0.28, p < .01), TWAKS (r = 0.23, p < .01), and TWAKO (r = 0.09, p < .05) indicating the inconsistency in sleep is positively associated with level of proces sing speed. Associations between within-person variability in sleep and mean-level sleep range d from 0.04 to 0.88, i ndicating that increased variability in sleep is associated with worse overa ll sleep. Associations between inconsistency in cognitive performance and level of performance i ndicated that both variability in reasoning and processing speed are associated with better reasoning (r = 0.34, p < .01) and better processing speed (r = 0.53, p < .01). Please see Table 5-4 fo r a complete listing of associations between within-person variability indi ces and mean-level indices ac ross sleep and cognition. Aim 3: To Determine the Between-Person (M ean-Level) and Within-Person (Day-to-DayLevel) Association between Sleep and Cogn ition in Older Adults with Insomnia Multicollinearity Prior to running the multilevel models to pred ict letter series and sy mbol digit performance a series of within-pers on and between-person corr elations was run among predictors variables to

PAGE 50

50 determine the extent to which the independent variables shared variances. Formal multicollinearity diagnostic procedures are not available for multilevel modeling. Within-person multicollinearity The within-person correlational analysis re vealed significant collinearity between SOLS and TWAKS (r = 0.10, p < .05), between WASOO and WASOS (r = 0.14, p < .01) and TWAKS (r = 0.11, p < .05) and SOLO (r = -0.10, p < .05), and between TWAKO and TWAKS (r = 0.16, p < .01) and SOLO (r = 0.15, p < .01) and WASOO (r = -0.12, p < .01). See Table 5-5 for a complete listing of within-person correlation estimates. Fo r a listing of within-p erson correlations among predictor variables when not cont rolling for time see Appendix D. Between-person multicollinearity The between-person correlati onal analysis revealed signi ficant collinearity between WASOS and SOLS (r = 0.48, p < .01), and between TWAKS and SOLS (r = 0.45, p < .05) and WASOS (r = 0.48, p < .01), and between SOLO and SOLS (r = 0.74, p < .01), and between TWAKO and SOLO (r = 0.44, p < .05) and WASOO (r = 0.50, p < .01). See Table 5-6 for a complete listing of between-person correlation estimates. For a listing of between-person correlations among predictor va riables when not controlli ng for time see Appendix D. Multilevel Model for Letter Series The intraclass correlatio n coefficient (ICC), which serves as an index of the amount of within and between-person vari ability to be explained (Br yk & Raudenbush, 1992), was 0.36. This indicates that, 64% of the overall variability in letter series was within-person and 36% was between-person. For a complete listing of model parameters and estimates obtained at each step of the model building process see Table 5-7. In the final MLM for letter series performance TWAKO was the only significant betweenperson, level 2, predictor, = 0.15, t (45.43) = 2.03, p = .05. At the within-person level, level 1,

PAGE 51

51 both the predictors of Day, = 0.37, t (46.87) = 8.53, p < .001, and SOLS, = 0.01, t (494.50) = 1.94, p = .05, were significant. A significant inte raction was found between level 2 and level 1 SOLS, = 0.00018, t (482.38) = -2.42, p < .05. The model also contained a significant random effect of Day, = 0.04, Walds Z = 2.73, p = .01. This model explained approximately 35% of the within-person vari ance and 25% of the between-person variance. The m odel accounted for roughly 86% of the total vari ance in letter series performance. See Table 5-8 for a total listing of predic tor estimates and significance levels. See Appendix E for total lis tings of predictor estimates a nd significance levels for MLMs estimates in model building steps 2-4. Multilevel Model for Symbol Digit The intraclass correlation coefficient (ICC) for symbol digit was 0.73. This indicates that, 27% of the overall variability in symbol digi t was within-person and 73% was between-person. For a complete listing of model parameters a nd estimates obtained at each step of the model building process see Table 5-9. In the final MLM for symbol digit performance, WASOS was the only significant between-person, level 2, predictor, = -0.08, t (1.47) = -2.08, p = .04. At the within-person level, level 1, only the predictor of Day, = 0.48, t (144.30) = 3.39, p = .001, was significant. A significant interaction was found be tween level 2 and level 1 TWAKS, = 0.001, t (465.84) = 2.13, p < .05, and level 2 and level 1 TWAK O, = 0.004, t (22.62) = 2.26, p < .05. The model contained no significan t random effects. This model explained approximately 1% of the within-person variance and 23% of the between-person variance. The m odel accounted for roughly 44% of the total varian ce in symbol digit performance performance. See Table 5-10 fo r a total listing of predictor estimates and

PAGE 52

52 significance levels. See Appendix F for total listings of predictor estimates and significance levels for MLMs estimates in model building steps 2-4.

PAGE 53

53 Table 5-1. Amount of within and between-person variability Sample Standard Deviation (SD) Individual Standard Deviation (ISD) SD / ISD Letter Series 4.96 2.38 0.48 Symbol Digit 9.05 8.36 0.92 Sleep Onset Latencys 31.58 26.86 0.85 Wake After Sleep Onsets 37.77 42.71 1.13 Terminal Wakefulnesss 25.03 31.32 1.25 Sleep Onset Latencyo 16.39 21.71 1.32 Wake After Sleep Onseto 13.09 14.72 1.12 Terminal Wakefulnesso 10.86 20.58 1.89 Notes: All values shown have been de-trended for any lin ear or quadratic effects of time. Variables with a subscript s indicate they were subjectively measure d by Sleep Diary. Variables with a subscript o indicate they were objectively measured by Actigraphy.

PAGE 54

54 0 0.5 1 1.5 2Le t t er Seri es Symbo l Di g it Sleep Onset Latency Wake After S l eep O nset T e rminal Wak e fulness Sleep O nset Latency W ake After S l eep On s e t T e rmin a l Wak e fulne s s Sleep Diary Actigraphy% of between-person variability Figure 5-1. Relative amount of w ithin-person variability compared to between -person variability after controlling for any linear or quadratic effects of time.

PAGE 55

55 Table 5-2. Correlations of within-p erson standard deviations (N = 48) 1 2 3 4 5 6 7 8 1. Letter Series 1.00 2. Symbol Digit 0.20** 1.00 3. Sleep Onset Latencys 0.06 0.40**1.00 4. Wake After Sleep Onsets -0.04 -0.06 0.45**1.00 5. Terminal Wakefulnesss 0.14** 0.35**0.53**0.28**1.00 6. Sleep Onset Latencyo 0.04 0.36**0.35**0.08* 0.14**1.00 7. Wake After Sleep Onseto 0.26** 0.31**0.12**0.24**0.17**0.42**1.00 8. Terminal Wakefulnesso 0.26** 0.37**0.31**0.30**0.26**0.33**0.62** 1.00 Note: ** Correlation is significant at the 0.01 level (2-t ailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript s indicate they were measured subjectively by Sleep Diary. Variables with a subscript o indicate they were objectively measured by Actigraphy. All values have been de-trended to control for any linear or quadratic effects of time.

PAGE 56

56 Table 5-3. Correlations of sample means (N = 48) 1 2 3 4 5 6 7 8 1. Letter Series 1.00 2. Symbol Digit 0.23** 1.00 3. Sleep Onset Latencys -0.14** 0.27** 1.00 4. Wake After Sleep Onsets -0.06 -0.06 0.43**1.00 5. Terminal Wakefulnesss -0.12** 0.22** 0.41**0.41**1.00 6. Sleep Onset Latencyo -0.09* 0.15** 0.61**0.09* 0.21**1.00 7. Wake After Sleep Onseto -0.16** 0.12** 0.14**0.22**0.18**0.04 1.00 8. Terminal Wakefulnesso 0.12** 0.18** 0.27**0.26**0.29**0.32** 0.32** 1.00 Notes: ** Correlation is significant at the 0.01 level (2 -tailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript s indicate they were subjectively measured by Sleep Diary. Variables with a subscript o indicate they were objectively measured by Actigraphy. All values have been de-trended to control for any linear or quadratic effects of time.

PAGE 57

57 Table 5-4. Correlations of w ithin-person standard deviations and sample means (N = 48) 1. (SD) 2. (SD) 3. (SD) 4. (SD) 5. (SD) 6. (SD) 7. (SD) 8. (SD) 1. Letter Series (M) 0.34** -0.15** -0.13** 0.00 -0 .24** -0.10* -0.07 -0.05 2. Symbol Digit (M) 0.34** 0.53** 0.28** -0.08 0.23** 0.02 0.00 0.09* 3. Sleep Onset Latency (M)s 0.24** 0.39** 0.77** 0.21** 0.45** 0.59** 0.29** 0.35** 4. Wake After Sleep Onset (M)s 0.09* 0.03 0.55** 0.60** 0.43** 0.13** 0.17** 0.22** 5. Terminal Wakefulness (M)s 0.30** 0.35** 0.22** 0.12** 0.69** 0.15** 0.15** 0.20** 6. Sleep Onset Latency (M)o 0.06 0.38** 0.30** 0.05 0.22** 0.88** 0.35** 0.28** 7. Wake After Sleep Onset (M)o 0.18** 0.24** 0.09* 0.04 0.24** 0.07 0.59** 0.32** 8. Terminal Wakefulness (M)o 0.21** 0.45** 0.23** 0.29** 0.25** 0.32** 0.58** 0.89** Notes: ** Correlation is significant at the 0.01 level (2-tailed) Correlation is significant at the 0.05 level (2-tailed). V ariables with a subscript s indicate they were subjectively measured by Sleep Diary. Variabl es with a subscript o indicate they were objectively measured by Actigraphy. All values have been de-trended to control for any linear or qua dratic effects of time. (M) indicates the value is the sample m ean. (SD) indicates the value is the within-person standard de viation. All values have been de-trended to control for any linear or quadratic effec ts of time.

PAGE 58

58 Table 5-5. Correlations of within-person predictors (N = 48, 14 Occasions) 1 2 3 4 5 6 7 8 1. Letter Series 1.00 2. Symbol Digit 0.08 1.00 3. Sleep Onset Latencys 0.00 0.00 1.00 4. Wake After Sleep Onsets 0.05 0.03 -0.04 1.00 5. Terminal Wakefulnesss 0.07 0.03 0.10* 0.05 1.00 6. Sleep Onset Latencyo -0.01 0.01 0.06 0.01 0.01 1.00 7. Wake After Sleep Onseto -0.01 0.04 0.08 0.14** 0.11** -0.10* 1.00 8. Terminal Wakefulnesso 0.07 0.04 -0.05 0.04 0.16** 0.15** -0.12** 1.00 Notes: ** Correlation is significant at the 0.01 level (2 -tailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript s indicate they were subjectively measured by Sleep Diary. Variables with a subscript o indicate they were objectively measured by Actigraphy. All values have been de-trended to control for any linear or quadratic effects of time.

PAGE 59

59 Table 5-6. Correlations of betw een-person predictors (N = 48). 1 2 3 4 5 6 7 8 1. Letter Series 1.00 2. Symbol Digit 0.23 1.00 3. Sleep Onset Latencys -0.14 0.30 1.00 4. Wake After Sleep Onsets -0.06 -0.07 0.48** 1.00 5. Terminal Wakefulnesss -0.15 0.25 0.45* 0.48** 1.00 6. Sleep Onset Latencyo -0.11 0.21 0.74** 0.11 0.29 1.00 7. Wake After Sleep Onseto -0.19 0.14 0.16 0.26 0.18 0.08 1.00 8. Terminal Wakefulnesso -0.14 0.25 0.41 0.34 0.39 0.44* 0.50** 1.00 Notes: ** Correlation is significant at the 0.01 level (2 -tailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript s indicate they were subjectively measured by Sleep Diary. Variables with a subscript o indicate they were objectively measured by Actigraphy. All values have been de-trended to control for any linear or quadratic effects of time.

PAGE 60

60 Table 5-7. Steps taken in building the Letter Series Multilevel Model Letter Series Models AIC BIC -2LL -2LL df df s2 b s2 w r2 b r2 w r2 t (1) Null 3226.303239.543220.30-3 -24.148.67--0.76 (2) Time added 3031.283053.363021.28199.02*** 4 1 23.766.590.020.240.82 (3) Sleep added 2755.962829.142721.96299.32*** 12 8 21.426.660.110.230.83 (4) Interactions added 2758.542857.542712.549.42 18 6 21.546.480.110.250.84 (5) Random effects added 2748.882865.102694.8817.66** 22 4 18.155.640.250.350.86 Notes: AIC = Akaikes Information Criterion; BIC = Schw arzs Bayesian Criterion; -2LL = -2 log likelihood; -2LL = change in LL relative to preceding model; s2 b = unexplained intercept-related (between subjects) variance; s2 w = unexplained residual-related (within subjects) variance; r2 b = between-subjects pseudo R-squared, an estimate of the amount of between subjects variance (estimated from null model) explai ned by fixed and random predictors; r2 w = within-subjects pseudo R-squared, an estimate of the amount of within subjects variance (estimated from null model) explained by fixed and random predictors; r2 t = total pseudo R-squared, an estimate of tota l variance explained (estimated from amount of agreement between predicted values and actua l values). *** Deviance is significant at the 0.001 level. ** Deviance is significa nt at the 0.01 level. Deviance is significant at the 0.05 level.

PAGE 61

61 Table 5-8. Sleep variables predicting reasoning Fixed Effects Predictor Variable B SE df t p Within-person Day 0.370.0446.878.53 <0.001Sleep Onset LatencyS 0.010.01494.501.94 0.05Wake After Sleep OnsetS 0.0030.005488.020.66 0.51Terminal WakefulnessS -0.0030.01490.64-0.46 0.65Sleep Onset LatencyS O -0.0040.01480.81-0.56 0.57Wake After Sleep OnsetO 0.010.02459.990.69 0.49Terminal WakefulnessO 0.010.01475.581.11 0.27Between-person Sleep Onset LatencyS 0.0020.0347.480.08 0.94Wake After Sleep OnsetS -0.0030.0246.43-0.14 0.89Terminal WakefulnessS -0.030.0347.29-0.90 0.37Sleep Onset LatencyS O -0.040.0546.77-0.69 0.50Wake After Sleep OnsetO -0.090.0646.29-1.56 0.13Terminal WakefulnessO 0.150.0745.432.03 0.05Interactions Level 1 Level 2 SOLS -0.000180.00007482.38-2.42 0.02Level 1 Level 2 WASOS -0.000020.00004471.81-0.54 0.59Level 1 Level 2 TWAKS 0.000180.00011499.911.61 0.11Level 1 Level 2 SOL O 0.000010.00013394.800.09 0.93Level 1 Level 2 WASO O -0.000330.00040455.47-0.82 0.41Level 1 Level 2 TWAK O -0.000260.00039457.58-0.67 0.51Random Effects Covariance parameter estimate B SE Z p Within-person Day 0.040.022.73 0.01Sleep Onset LatencyS 0.000.000.00 1.00Level 1 Level 2 SOLS 0.000.000.00 1.00Level 1 Level 2 TWAKS 0.000.000.00 1.00 Within Pseudo R2 0.35 Between Pseudo R2 0.25 Total Pseudo R2 0.86 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 62

62 Table 5-9. Steps taken in building the Symbol Digit Multilevel Model Symbol Digit Models AIC BIC -2LL -2LL df df s2 b s2 w r2 b r2 w r2 t (1) Null 4498.874511.944492.87-3 -68.42121.52--0.39 (2) Time added 4471.834493.624461.8331.04*** 4 1 63.91121.970.070.000.41 (3) Sleep added 4021.514093.633987.51474.32*** 12 8 52.18123.000.24-0.010.43 (4) Interactions added 4018.754116.323972.7514.76* 18 6 52.50120.280.230.010.44 (5) Random effects added 4026.594141.133972.590.16 22 4 52.41120.030.230.010.44 Notes: AIC = Akaikes Information Criterion; BIC = Schw arzs Bayesian Criterion; -2LL = -2 log likelihood; -2LL = change in LL relative to preceding model; s2 b = unexplained intercept-related (between subjects) variance; s2 w = unexplained residual-related (within subjects) variance; r2 b = between-subjects pseudo R-squared, an estimate of the amount of between subjects variance (estimated from null model) explai ned by fixed and random predictors; r2 w = within-subjects pseudo R-squared, an estimate of the amount of within subjects variance (estimated from null model) explained by fixed and random predictors; r2 t = total pseudo R-squared, an estimate of tota l variance explained (estimated from amount of agreement between predicted values and actua l values). *** Deviance is significant at the 0.001 level. ** Deviance is significa nt at the 0.01 level. Deviance is significant at the 0.05 level.

PAGE 63

63 Table 5-10. Sleep variables predicting processing speed Fixed Effects Predictor Variable B SE df t p Within-person Day 0.48 0.14 144.30 3.39 0.001Sleep Onset LatencyS -0.02 0.03 461.39 -0.56 0.58Wake After Sleep OnsetS -0.02 0.02 459.24 -1.15 0.25Terminal WakefulnessS -0.04 0.03 446.96 -1.58 0.12Sleep Onset LatencyS O 0.03 0.03 442.21 0.82 0.41Wake After Sleep OnsetO 0.14 0.08 421.86 1.82 0.07Terminal WakefulnessO -0.11 0.06 95.13 -1.91 0.06Between-person Sleep Onset LatencyS 0.11 0.06 43.77 1.83 0.07Wake After Sleep OnsetS -0.08 0.04 41.47 -2.08 0.04Terminal WakefulnessS 0.07 0.06 43.36 1.14 0.26Sleep Onset LatencyS O -0.07 0.10 44.58 -0.71 0.48Wake After Sleep OnsetO 0.05 0.10 41.94 0.49 0.62Terminal WakefulnessO 0.09 0.14 42.52 0.69 0.49Interactions Level 1 Level 2 SOLS 0.0002 0.0003 458.90 0.52 0.61Level 1 Level 2 WASOS 0.0002 0.0002 430.17 0.98 0.33Level 1 Level 2 TWAKS 0.0010 0.0005 465.84 2.13 0.03Level 1 Level 2 SOL O -0.0005 0.0006 380.48 -0.90 0.37Level 1 Level 2 WASO O -0.0028 0.0018 390.40 -1.60 0.11Level 1 Level 2 TWAK O 0.0042 0.0019 22.62 2.26 0.03Random Effects Covariance parameter estimate B SE Z p Within-person Day 0.00 0.000.00 1.00Terminal WakefulnessO 0.00 0.000.00 1.00Level 1 Level 2 TWAKS 0.00 0.000.00 1.00Level 1 Level 2 TWAK O 0.000001 0.0000020.35 0.72 Within Pseudo R2 0.01 Between Pseudo R2 0.23 Total Pseudo R2 0.44 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 64

64 CHAPTER 6 DISCUSSION This section has four main objectives: to revi ew and interpre t the results of each primary aim of the study; to consider the major limitations of the study; to discuss the larger theoretical and empirical implications; and to discuss futu re directions for the study of the sleep-cognition relationship. Review of Findings Aim 1: To Determine the Amount of Variab ility in Sleep and Cognition Found Within Older Adults with Insomnia Compared to the Amount of Variability between Older Adults with Insomnia Results from this analysis indicate that a substantial amount of va riability in sleep and cognition is found within older adults with insomn ia after controlling fo r any effects of time. Variability in the letter series test was the lo west of all the variab les measured and still considerable (48%). Every other variable asse ssed exhibited either e qual amounts of withinperson variability as between-person variability or more (range = 85% 189%). Results suggest that within-person variability in sleep and cogniti on in older adults with insomnia is large enough to warrant further investigation. Results are congruent with the literature in both the sleep and cognitive field demonstrating large amounts of within-person variability in their relative constructs of study (Allaire & Ma rsiske, 2005; Coates et al., 1981; Edinger et al., 1991; Edinger et al., 1997; Frankel et al., 1976; Hauri & Wisbey 1992; Hultsch et al., 2000; Hultsch et al., 2002; MacDonald et al., 2003; Ne sselroade & Salthouse, 2004; Sa lthouse et al., 2006; Salthouse & Berish, 2005 Vallieres et al., 2005) Aim 2: To Determine How Variability is Asso ciated Across Sleep and Cognitive Measures in Older Adults with Insomnia The first question this aim a ddressed was how intraindividual variability is related across the various sleep and cognition domains. Results indicated that variability in one domain is

PAGE 65

65 related to variability in other domains. It appears that variability may operate systematically within-persons. Individual s who operate in a variable manner in one task also operate variably in other tasks. These findings are inline with the bulk of previous studi es of intraindividual variability (Allaire & Marsiske, 2005; Hultsch et al., 2000; Nesselroa de & Salthouse, 2004). Interestingly, variability in WASOS was negatively, though non-signi ficantly, correlated with variability in cognition. The second question this aim addressed was how level of performance is related across the various sleep and cognition domains. Interestingly, results indicate that in general better sleep is related to lower letter seri es performance and higher sy mbol digit scores. Although not definitive, this pattern of findings supplies some evidence for the notion that the symbol digit test may be more indicative of unde rlying brain functions than the letter series test. The third question this aim addressed was how variability is related to level of performance across the various sleep and cognition measures. Increa sed variability in both letter series and symbol digit were related to increa sed overall level of sl eep. This suggests that individuals who had higher mean values on the sleep variables had more variable cognitive performance. In light of the fact that all of th e sleep parameters assessed are typically used as indicators of poor sleep and assu ming more variability in cogni tion is a bad thing, this finding make sense. If variability is indicative of underlying neuronal dysfunction (Sliwinski, Hofer, & Hall, 2003), this result lends support to the notion that sleep provides a restorative period for the brain. Increased variability in sleep parameters was generally negatively associated with average performance on the letter series task and positively associated with average symbol digit score. Although seemingly contradictory in indications, these results do fit prior research findings suggesting that variability may be adaptive (a good thing) or a sign of neural deterioration (a bad

PAGE 66

66 thing) (Allaire & Marsiske, 2005). Increased variability in slee p was associated with poorer overall sleep. Variable sleep is typically viewed as an indicati on of poor sleep, as suggested by the findings that insomniacs tend to be more va riable in their sleep patterns than normal good sleepers (Coates et al., 1981; Edinge r et al., 1997). This set of results confirms that more variable sleep is associated with poorer sleep. Increased variability in both reasoning and processing speed were found to be associated with better reasoning and processing speed. These results add to the current literature suggesti ng that variability may be adaptive in nature (Allai re & Marsiske, 2005; Li, Aggen, Nesselroade, & Baltes, 2001; Siegler, 1994). Aim 3: To Determine the Between-Person (M ean-Level) and Within-Person (Day-to-DayLevel) Association between Sleep and Cogn ition in Older Adults with Insomnia An initial correlational analysis of predicto r variables, both within and between-persons, revealed that no two within-person predictors shar ed more than 2.5% of their variance and that no two between-person predictors shared more than 55% of their variance. MLM for reasoning The intraclass correlation for le tter series was 0.36 indicting that 64% of the variance in letter series was within-persons and investigation of this within-p erson variability was warranted. After controlling for time, within-person variations in SOLS significantly predicted letter series performance. Results sugge st that on days when an indivi dual self-estimates taking longer to fall asleep, they also have better reasoning performance. A significant positive between-person association was found between TWAKO and letter series indicati ng that individuals who, on average, tend to spend more time laying in bed in the morning also have better than average reasoning. TWAK, in general, has been an unders tudied variable. Result s that increased TWAK may be beneficial to cognitive performance could indicate an attempt to compensate for a poor

PAGE 67

67 nights sleep that works. Yet, TWAK is largely considered maladaptive in sleep research. Thus, additional research is needed before definitive conclusions can be drawn. There was also a significant negative inte raction between within-person and betweenperson SOLS indicating that when an individual who usually has above average SOLS experiences a day of increased SOLS they perform worse on the reasoning task. Though these results may seem puzzling at first they are very much inline with curr ent findings on the sleepcognition relationship. The fact that when individuals experience above average SOLS and when individuals tend to, on average, have more TWAKO they also perform better on the letter series could be evidence for the hypothesized added effort such individu als are suspected of expending to offset the natural decline following poor sleep (Drake et al., 2001). Furthermore, when individuals who usually take a long time to fall asleep take even longer they consequently experience lower than average reasoning. This could be interpreted as creating too much impairment for added effort to compensate for, as in total sleep deprivation. This model accounted for 35% of the within-person vari ance in letter seri es and 25% of the between-person variance. In total, the model accounted for 86% of the variance in reasoning performance. MLM for processing speed The intraclass correlation for le tter series was 0.73 indicting that 27% of the variance in letter series was within-persons and investigation of this within-p erson variability was warranted. After controlling for time there were no significant within-per son predictors of symbol digit performance. A negative significant between-person association between WASOS and symbol digit was found. This rela tionship indicates that individua ls who, on average, spend more self-reported wake time in the middle of the ni ght also have lower than average processing

PAGE 68

68 speeds. This finding is consistent with the sl eep deprivation research ing showing a negative relationship between sleep loss a nd cognitive performance in olde r adults (Webb, 1985; Web & Levy, 1982). Two significant interactions were found. The within-person by between person interaction for both TWAKS and TWAKO were positively related to symbol digit performance. These interactions indicate that when an individual who usually has above average TWAK experiences a day of increased TWAK they have better pr ocessing speeds. As stated above, TWAK is a recently introduced variable into th e sleep literature. These interac tions suggest that getting much more rest time in the morning may supply a type of restorative function fo r the prefrontal cortex that was not achieved at night (Horne, 1988). Increasing TWAK may be an adaptive behavior to the loss of nocturnal sleep in or der to maintain optimal cognitive functioning. However, as stated previously, TWAK is generally agreed upon to be a maladaptive sleep response and thus needs further research before definitive conclusions are drawn. This model accounted for only 1% of the within -person variance in symbol digit indicating the need for better within-person predictors. However, the model accounted for 23% of the between-person variance and 44% of the total variance in processing speed. Study Limitations There are several important study limitations that need to be addressed. While the data presented here were well suited to answer the questions asked; this is a secondary data analysis. Therefore, it is important to recognize that the da ta were not collected to answer the questions of within-person variab ility in sleep and cognition over 14 da ys in older adults with insomnia. However, it also must be recognized that the da ta were collected for a closely related purpose and as such do fit the current uses extremely well.

PAGE 69

69 The lack of a normal sleeping control group restricts the interpre tation of results. The presence of such a comparison would have al lowed statements about the differing magnitude, differing associations, and differing daily couplin g of sleep and cognition. As is, the current study is confined to addressing these importa nt questions in older insomniacs only. Generalizability of results may be further hinde red by the sample characteristics of the study. Specifically, study participants were above-ave rage educated and mostly Caucasian. Lastly, the use of self-administered daily assessments and the resulting loss of experimental control must be acknowledged. Pa rticipants were thoroughly instructed in the procedures of all material and we re provided with all necessary ma terial and then instructed to complete everything once daily. Although the possibi lity of data integrity problems definitely exists, we believe this is highly unlikely for seve ral reasons. A very simila r protocol was used in a previous study of intraindivi dual variability suc cessfully (Allaire & Marsiske, 2005). Selfassessments were also monitored weekly to decr ease the probability of incorrect administration. And, all workbooks contained integrity statemen ts that were all signed and returned. Theoretical and Empirical Implications With these limitations in mind, we feel this study has several importa nt implications for sleep researchers, cognitive aging researchers, and real world applications First, this research reveals the importance of within-per son variations in sleep. Intrai ndividual variability has yet to become considered a salient topic in sleep research. Instead many researchers are focusing exclusively on interindividual variab ility at the expense of intraindi vidual variability. Results of the current study indicate that with in-person variations in sleep ar e not noise or measurement error and suggest the need to systematically stud y how individuals sleep varies from day-to-day. Specifically, the results of Aim 3 illustrate th at disregarding within-person variability as measurement error would lead to decreased pr ecision in predicting cognition with sleep.

PAGE 70

70 It is also hoped that the resu lts of this study will inspire mo re sleep researchers to become familiar with multilevel modeling and its applic ation for studying within-person variability. Several researchers have previously employe d multilevel modeling bu t failed to decompose predictors into within an d between components (Monk et al., 2006). By doing this these researchers may have missed possibly intrigui ng results about within -person processes. Furthermore, several researchers have commented on the possibility that the inconsistencies in the results of sleep deprivation studies are due to stable interindi vidual differences in susceptibility to sleep loss (Olofsen et al., 2004 ; Van Dongen et al., 2004). It is our contention that through the use of MLM and testing random effects such hypotheses can be tested. This study adds to the sleep research literat ure an investigation of the sleep-cognition relationship that is ecologically valid. While ma ny studies have deprived people of their sleep and then tested their cognitive functioning (Har rison & Horne, 2000; Pilcher & Huffcutt, 1996) very few have attempted to capture this re lationship in an already poor sleeping sample. Furthermore, those studies that did assess individu als in their natural setti ng settled for relatively brief cognitive measures instead of commonly used, domain specific assessment (Blackwell et al., 2006; Cricco et al., 2001; Twor oger et al., 2006). Aim 3 also poi nts for the need of increased research into the utility of TWAK as a sleep va riable. TWAK is traditionally viewed as an unrecommended response to poor nocturnal sleep and as a contributor to subs equent nights of poor sleep. Our results suggest that TWAK may be bene ficial to cognitive performance in older adults with insomnia. We suggest future research examine this relationship in more detail. In this way, the current study adds depth to the current stat e of sleep research examining the sleep-cognition link.

PAGE 71

71 Implications for the cognitive aging field are straightforward; sleep is associated with cognitive performance in older adults. Given the known changes in sleep that occurs with increased age (Morgan, 2000) and the known relati onship between sleep a nd cognition (Harrison & Horne, 2000; Pilcher & Huffcutt, 1996) it is essential that sleep be more readily studied within the cognitive aging field. Is it coincidence that executive functioning, which is regulated by the prefrontal cortex, has been found to exhibit high levels of variability (S althouse et al., 2006) and that the prefrontal cortex is the most sensit ive region of the brain to sleep loss (Durmer & Dinges, 2005) and that sleep exhibi ts much variability in normal sleepers and older adults with insomnia (Coates et al., 1981; Edinger et al., 199 1; Edinger et al., 1997; Frankel et al., 1976; Hauri & Wisbey, 1992; Vallieres et al., 2005; McCrae et al., 2003; McCrae et al., 2005)? Implications of the current study for real worl d applicability are many. First, a reliable relationship was illustrated between sleep and cogn itive functioning. It is therefore important for individuals doing cognitive/neurops ychological assessments of older adults to also assess sleep. Given the association between sl eep loss and accidents, auto a nd work related, (Benca, 2001; Hublin & Partinen, 2002; Katz & McHorney, 2002; Neubauer, 2004; Roth & Roehrs, 2003) and the daily associations between sleep and percep tual speed found in the current study, it is imperative to warn older drivers of the potential negative consequences of one nights bad sleep. There is a general cognitive de cline experienced with increas ing age (Craik & Byrd, 1982; Hasher & Zacks, 1988; Lindenberger & Baltes, 1 994; Salthouse, 1996). Ther e is also a general worsening of sleep associated with advanced age (Morgan, 2000). If these two phenomena are causally related then a possible intervention to help slow one possible factor in age-related cognitive decline could be treatments aimed at improving sleep. Future research should investigate the longitudinal effects on c ognitive performance of improved sleep.

PAGE 72

72 Future Directions Possible follow-up studies to the current project could proceed in many different directions. Given the possible re lationship between REM sleep and memory consolidation (Gais & Born, 2004; Wagner et al., 2001) and facilitati on (Walker & Stickgold, 2004) a potential line of research would be to have individuals wear portable polysomnographies equipped with electroencephalograms nightly a nd measure word recall daily to determine the daily association between amount of REM sleep obtain ed and memory performance. Similar studies as the current one could be conducted that included normal controls for comparison purposes. Further, future studies shou ld examine more occasions, in a larger more diverse sample, and should assess more cognitiv e domains using more pr ecise tools, such as computers, to allow for an accurate recordi ng of individual reacti on times. Such innovation would allow for the examination of not only oc casion-to-occasion variability but also within occasion variability. Another possible vein of research would be the detection of other covariates that could be incorporated into the MLM to aid in the predic tion of cognitive performance. Possible covariates might include: affect, physical act ivity, life stressors, and pain.

PAGE 73

73 APPENDIX A SLEEP DIARY Please answer the following questionnaire WHEN YOU AWAKE IN THE MORNING. Enter yesterday's day and date and provide the informa tion to describe your sleep the night before. Definitions explaining each line of the questionnaire are given below. Table A-1. Example of a Sleep Diary yesterday's day yesterday's date TUES 10/14/9 7 day 1 day 2 day 3 day 4 day 5 day 6 day 7 1. NAP (yesterday) 70 2. BEDTIME (last night) 10:55 3. TIME TO FALL ASLEEP 65 4. # AWAKENINGS 4 5. WAKE TIME (middle of night) 110 6. FINAL WAKE-UP 6:05 7. OUT OF BED 7:10 8. QUALITY RATING* 2 9. BEDTIME MEDICATION (include amount & time) Halcion 0.25 mg 10:40 pm *Pick one number below to indicate your overall QUALITY RATING or satisfaction with your sleep.1. very poor, 2. poor, 3. fair, 4. good, 5. excellent

PAGE 74

74 APPENDIX B LETTER SERIES TEST Circle the one letter from the answer choices on the right that comes next in the series of letters on the left. Answer Choices 1. o o p q q r s s t u u o q t u v 2. o p q o p q r s t r s t u v w u v w x y 3. o s o t p s p t q s q t r s r p q r s t 4. f g h p f g h q f g h r f g h f q r s t 5. n o p i j k q r s i j k t u v w x y i j 6. u v w u v x u v y u v z u v w y z 7. d e p f g p h i p j k p l m p d e l m n 8. q p o q p o q p o q p o p q r s 9. p a p q a p q r a p q r s r s t u a 10. s t s t p q u v u v p q w x p q w x y 11. o p p q q q r r r r s s s s r s t u v 12. o o q q s s u u w w v w x y z 13. r s t r s t u r s t u v r s t u v 14. o p q q r s t t u v w w x y z v w x y z

PAGE 75

75 Answer Choices 15. i j i k l k m n m o p o q r o p q r s 16. i j k v w l m n v w o p q v w q r s v w 17. i j j j k l l l m n n n o p p o p q r s 18. p o n m l k j i j k o p 19. i k m o q s u s t u v w 20. i j j k l l m n n o p m n o p q 21. i p j q k r i p j q k r i p j i j k p q 22. h i k l n o q r p q r s t 23. i j k i l m n l o p q o r s t q r s t u 24. i a j b k c l d m e n f o n o p f g 25. a c f h k m p r q r s t u 26. i i j j k l l m m n o o p p q r s t 27. i i j i j k k l k l m m n m n o p q 28. n n n n n o o o o p p p q m n o p q 29. i j k k j i l m n n m l o p q p q r s t 30. i j k j k l m n m n o p q p n o p q r

PAGE 76

76 APPENDIX C SYMBOL DIGIT TEST On the next page are boxes to be filled in with the matching digit, like you see in the key on the top of the page. Go in order, without skipping any boxes. KEY 1 2 3 4 5 6 7 8 9 4 1 2 9 4 5 2 7 1 6

PAGE 77

77

PAGE 78

78 APPENDIX D CORRELATIONS OF PREDICTOR VARI ABLES NOT CONTROLLING FOR TIME Table D-1. Correlations of within-person, with time, predictors. (N=48, 14 Occasions) 1 2 3 4 5 6 7 8 1. Letter Series 1.00 2. Symbol Digit 0.16** 1.00 3. Sleep Onset Latency 0.01 0.01 1.00 4. Wake After Sleep Onset 0.03 0.02 -0.05 1.00 5. Terminal Wakefulness 0.05 0.03 0.10** 0.05 1.00 6. Sleep Onset Latency -0.02 0.00 0.06 0.01 0.01 1.00 7. Wake After Sleep Onset 0.03 0.05 0.08 0.14** 0.11** -0.10* 1.00 8. Terminal Wakefulness 0.06 0.04 -0.05 -0.05 0. 15** 0.15** -0.12** 1.00 Note: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript s indicate they were subjectively measured by Sleep Diary. Variables with a subscript o i ndicate they were objectively measured by Actigraphy. Table D-2. Correlations of between-per son, with time, predictors. (N=48) 1 2 3 4 5 6 7 8 1. Letter Series 1.00 2. Symbol Digit 0.23 1.00 3. Sleep Onset Latency -0.14 0.30 1.00 4. Wake After Sleep Onset -0.06 -0.07 0.48** 1.00 5. Terminal Wakefulness -0.15 0.25 0.45* 0.48** 1.00 6. Sleep Onset Latency -0.11 0.21 0.74** 0.11 0.29 1.00 7. Wake After Sleep Onset -0.18 0.13 0.16 0.26 0.18 0.08 1.00 8. Terminal Wakefulness 0.14 0.26 0.41 0.34 0.41 0.45* 0.50** 1.00 Note: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript s indicate they were subjectively measured by Sleep Diary. Variables with a subscript o i ndicate they were subjectively measured by Actigraphy.

PAGE 79

79 APPENDIX E REASONING MODELS AT EACH STEP OF MLM BUILDING Table E-1. Reasoning ML M Step 2: Adding Time Fixed Effects Predictor Variable B SE df t p Within-person Day 0.380.03148.2912.40 <0.00Between-person ------Interactions ------Random Effects Covariance parameter estimate B SE Z p Within-person -----Within Pseudo R2 0.24 Between Pseudo R2 0.02 Total Pseudo R2 0.82 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 80

80 Table E-2. Reasoning ML M Step 3: Adding Sleep Fixed Effects Predictor Variable B SE df t p Within-person Day 0.370.03140.2911.42 <0.00Sleep Onset LatencyS 0.000.00494.240.05 0.96Wake After Sleep OnsetS 0.000.00481.360.19 0.85Terminal WakefulnessS 0.010.00498.631.73 0.08Sleep Onset LatencyS O 0.000.00439.32-0.69 0.49Wake After Sleep OnsetO 0.000.01489.10-0.06 0.95Terminal WakefulnessO 0.010.00477.541.12 0.26Between-person Sleep Onset LatencyS -0.010.0345.74-0.29 0.77Wake After Sleep OnsetS 0.010.0245.370.28 0.78Terminal WakefulnessS -0.030.0345.69-0.92 0.36Sleep Onset LatencyS O -0.030.0645.87-0.56 0.58Wake After Sleep OnsetO -0.090.0645.39-1.58 0.12Terminal WakefulnessO 0.130.0845.481.73 0.09Interactions ------Random Effects Covariance parameter estimate B SE Z p Within-person -----Within Pseudo R2 0.23 Between Pseudo R2 0.11 Total Pseudo R2 0.83 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 81

81 Table E-3. Reasoning MLM St ep 4: Adding Interactions Fixed Effects Predictor Variable B SE df t p Within-person Day 0.370.03141.8611.57 <0.00Sleep Onset LatencyS 0.010.01497.631.97 0.05Wake After Sleep OnsetS 0.000.00488.990.46 0.64Terminal WakefulnessS 0.000.01484.46-0.53 0.60Sleep Onset LatencyS O 0.000.01474.18-0.45 0.65Wake After Sleep OnsetO 0.010.02452.800.53 0.59Terminal WakefulnessO 0.010.01462.950.89 0.38Between-person Sleep Onset LatencyS -0.010.0345.72-0.29 0.77Wake After Sleep OnsetS 0.010.0245.350.24 0.81Terminal WakefulnessS -0.030.0345.65-0.90 0.37Sleep Onset LatencyS O -0.030.0645.84-0.57 0.57Wake After Sleep OnsetO -0.090.0645.38-1.57 0.12Terminal WakefulnessO 0.130.0845.451.73 0.09Interactions Level 1 Level 2 SOLS 0.000.00496.60-2.40 0.02Level 1 Level 2 WASOS 0.000.00459.20-0.46 0.65Level 1 Level 2 TWAKS 0.000.00497.891.99 0.05Level 1 Level 2 SOL O 0.000.00397.25-0.01 0.99Level 1 Level 2 WASO O 0.000.00453.78-0.69 0.49Level 1 Level 2 TWAK O 0.000.00440.46-0.52 0.61Random Effects Covariance parameter estimate B SE Z p Within-person -----Within Pseudo R2 0.25 Between Pseudo R2 0.11 Total Pseudo R2 0.84 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 82

82 APPENDIX F PROCESSING SPEED MODELS AT EACH STEP OF MLM BUILDING Table F-1. Processing Speed MLM Step 2: Adding Time Fixed Effects Predictor Variable B SE df t p Within-person Day 0.480.13160.443.59 <0.00Between-person ------Interactions ------Random Effects Covariance parameter estimate B SE Z p Within-person -----Within Pseudo R2 0.00 Between Pseudo R2 0.07 Total Pseudo R2 0.41 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 83

83 Table F-2. Processing Speed MLM Step 3: Adding Sleep Fixed Effects Predictor Variable B SE df t p Within-person Day 0.470.14148.903.36 <0.00Sleep Onset LatencyS 0.000.01470.82-0.27 0.79Wake After Sleep OnsetS 0.000.01456.11-0.25 0.80Terminal WakefulnessS 0.010.01469.900.49 0.62Sleep Onset LatencyS O 0.000.02414.490.05 0.96Wake After Sleep OnsetO 0.020.03468.200.72 0.47Terminal WakefulnessO 0.020.02451.040.98 0.33Between-person Sleep Onset LatencyS 0.100.0643.651.80 0.08Wake After Sleep OnsetS -0.080.0441.32-2.05 0.05Terminal WakefulnessS 0.070.0643.261.14 0.26Sleep Onset LatencyS O -0.070.1044.48-0.65 0.52Wake After Sleep OnsetO 0.050.1041.850.52 0.61Terminal WakefulnessO 0.100.1442.420.71 0.48Interactions ------Random Effects Covariance parameter estimate B SE Z p Within-person -----Within Pseudo R2 -0.01 Between Pseudo R2 0.24 Total Pseudo R2 0.43 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 84

84 Table F-3. Processing Speed ML M Step 4: Adding Interactions Fixed Effects Predictor Variable B SE df t p Within-person Day 0.480.14146.193.36 <0.00Sleep Onset LatencyS -0.020.03465.87-0.56 0.58Wake After Sleep OnsetS -0.020.02459.51-1.10 0.27Terminal WakefulnessS -0.040.03455.71-1.57 0.12Sleep Onset LatencyS O 0.030.03444.730.80 0.43Wake After Sleep OnsetO 0.150.08437.491.89 0.06Terminal WakefulnessO -0.110.06428.38-2.03 0.04Between-person Sleep Onset LatencyS 0.110.0643.761.84 0.07Wake After Sleep OnsetS -0.080.0441.46-2.08 0.04Terminal WakefulnessS 0.070.0643.351.14 0.26Sleep Onset LatencyS O -0.070.1044.57-0.71 0.48Wake After Sleep OnsetO 0.050.1041.940.49 0.62Terminal WakefulnessO 0.090.1442.510.69 0.49Interactions Level 1 Level 2 SOLS 0.000.00465.480.51 0.61Level 1 Level 2 WASOS 0.000.00433.480.96 0.34Level 1 Level 2 TWAKS 0.000.00467.092.14 0.03Level 1 Level 2 SOL O 0.000.00382.83-0.87 0.38Level 1 Level 2 WASO O 0.000.00429.51-1.70 0.09Level 1 Level 2 TWAK O 0.000.00410.502.51 0.01Random Effects Covariance parameter estimate B SE Z p Within-person -----Within Pseudo R2 0.01 Between Pseudo R2 0.23 Total Pseudo R2 0.44 Notes: Variables with a subscript s indicate th ey were subjectively measured by Sleep Diary. Variables with a subscript o indicate th ey were objectively measured by Actigraphy.

PAGE 85

85 LIST OF REFERENCES Allaire, J. C., & Marsiske, M. (2005). Intrai ndividual variability may not always indicate vulnerability in elders cognitive performance. Psychology and Aging 20, 390-401. American Academy of Sleep Medicine. (2005). The International Classification of Sleep Disorders: Diagnostic and Coding Manual, 2nd Edition Westchester, IL: American Academy of Sleep Medicine. American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders Washington, DC: Author. Bastein, C. H., Fortier-Brochu, E., Rioux, I., LeBlanc, M., Dale y, M., & Morin, C. M. (2003). Cognitive performance and sleep quality in the elderly suffering from chronic insomnia: Relationship between objective and subjective measures. Journal of Psychosomatic Research 54, 39-49. Beck, A. T., Brown, G., & Steer, R. A. (1996). Beck Depression Inventory II manual San Antonio, TX: The Psychological Corporation. Belenky, G., Wesenten, N. J., Thorne, D. R., T homas, M. L., Sing, H C., Redmond, D P., Russo, M. C., & Balkin, T. J. (2003). Patterns of performance degradation and restoration during sleep restriction and subsequent r ecovery: a sleep dose-response study. Journal of Sleep Research 12, 1-12. Benca, R. M. (2002). Consequences of insomnia and its therapies. Journal of Clinical Psychiatry 62, 33-38. Blackwell, T., Yaffe, K., Ancoli-Isreal, S., Schneider J. L., Cauley, J. A., Hiller, T. A., Fink, H. A., & Stone, K. L. (2006). Poor sleep is asso ciated with impaired cognitive function in older women: The study of osteoporotic fractures. Journal of Gerontology 61, 405-410. Bliwise, D. (1993). Sleep in normal aging and dementia. Sleep 16, 40-81. Bosselli, M., Parrino, L., Smerieri, A., & Terza no, M. G. (1998). Effects of age on the EEG arousals in normal sleep. Sleep 21, 351-357. Bryk, A. S. & Raudenbush, S. W. (1992). Hierarchical linear models : Applications and data analysis methods Thousand Oaks: Sage. Burton, C. L., Hultsch, D. F., Strauss, E., & Hunt er, M. A. (2002). Intraindividual variability in physical and emotional functioni ng: Comparison of adults with traumatic brain injury and healthy adults. The Clinical Neuropsychologist 16, 264-279. Butler, A. C., Hokanson, J. E., & Flynn, H. A ., (1994). A comparison of self-esteem lability and low trait self-esteem as vulnerability factors for depression. Journal of Personality and Social Psychology 66, 166-177.

PAGE 86

86 Craik, F. I. M., & Byrd, M. (1982). Aging a nd cognitive deficits: The role of attentional resources. In F. I. M. Craik & S. Trehub (Eds.), Aging and cognitive processes New York: Plenum Press. Coates, T. J., George, J. M., Killen, J. D., Ma rchini, E., Hamilton, S. and Thorensen, C. E. (1981). First night effects in good sleepers and sleep-maintenance insomniacs when recorder from home. Sleep, 4, 293-8. Crenshaw, M. C., & Edinger, J. D. (1999). Sl ow-wave sleep and waking cognitive performance among older adults with and without insomnia complaints. Physiology & Behavior 66, 485-492. Cricco, M., Simonsick, E. M., & Foley, D. J. (2001). The impact of insomnia on cognitive functioning in older adults. Journal of the American Geriatrics Society 49, 1185-1189. Dinges, D. F., Pack, F., Williams, K., Gillen, N. A., Powell, J W., Ott, G E., Aptowicz, C., & Pack, A. I. (1997). Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4-5 hours per night. Sleep 20, 267-277. Drake, C. L., Roehrs, T. A., Burduvali, E., Bonahoom, A., Rosekind, M., & Roth, T. (2001). Effects of rapid versus slow accumu lation of eight hours of sleep loss. Psychophysiology 38, 979-987. Drummond, S. P. A., Brown, G. A., Gillin, J. C., Stricker, J. L ., Wong, E. C., & Buxton, R. B. (2000). Altered brain response to verbal learning following sleep deprivation. Nature 403, 655-657. Durmer, J. S., & Dinges, D. F. (2005). Neuroc ognitive consequences of sleep deprivation. Seminars in Neurology 25, 117-129. Edinger, J. D., Fins, A. I., Sullivan, R. J., Jr., Marsh, G. R., Dailey, D. S., Hope, T. V., Young, M., Shaw, E., Carlson, D. and Vasilas, D. (1997). Sleep in the la boratory and sleep at home: Comparisons of older inso mniacs and normal sleepers. Sleep, 20, 1119-26. Edinger, J. D., Marsh, G. R., Mccall, W. V., Er win, C. W. and Lininger, A. W. (1991). Sleep variability across consecutive nights of home monitoring in older mixed DIMS patients. Sleep, 14, 13-17. Eizenman, D. R., Nesselroade, J. R., Featherman D. L., & Rowe, J. W. (1997). Intraindividual variability in perceived cont rol in an older sample: The McArthur Successful Aging Studies. Psychology and Aging 12, 489-502. Espie, C. A. (1991) The psychological treatment of insomnia. Chichesster, UK: J. Wiley and Sons.

PAGE 87

87 Floyd, J. A., Medler, S. M., Ager, J. W., & Jani sse, J. J. (2000). Age -related changes in the initiation and maintenance of sleep: A meta-analysis. Research in Nursing and Health 23, 106-117. Folstein, M. F., Folstein, S. E. & McHugh, P. R. (1975). Mini Ment al State: A practical method for grading the cognitive states of patients for the clinician. Journal of Psychiatric Research, 12 189-198. Frankel, B. L., Coursey, R. D., Buchbinder, R. and Snyder, F. (1976). Recorded and reported sleep in chronic primary insomnia. Archives of General Psychiatry, 33, 615-23. Gais, S., & Born, J. (2004). Low acetycholine duri ng slow-wave sleep is cr itical for declarative memory consolidation. Proceedings of the National Academy of Science 101, 2140-2144. Harrison, Y., & Horne, J. A. (2000). The imp act of sleep deprivation on decision making: A review. Journal of Experimental Psychology 6, 236-249. Hasher, L., & Zacks, R. T., (1988). Working memo ry, comprehension and aging: A review and a new view. In G. H. Bower (ed.), The psychology of le arning and motivation San Diego: Academic Press. Hauri, P. J. and Wisbey, J. (1992). Actigraphy and insomnia: A closer look. Sleep, 15, 293-301. Hobson, J. A., & Pace-Schott, E. F. (2002). The cognitive neuroscience of sleep: Neuronal systems, consciousness and learning. Nature Reviews 3, 679-693. Horne, J. A. (1988). Sleep depriva tion and divergent thinking ability. Sleep 11, 528-536. Horne, J. A. (1993). Human sleep, sleep depr ivation and behaviour: Implications for the prefrontal cortex and psychiatric disorder. British Journal of Psychiatry 162, 413-419. Hublin, C. G. M., & Partinen, M. M. (2002). Th e extent and impact of insomnia as a public health problem. Journal of Clinical Psychiatry 4, 8-12. Hultsch, D. F., MacDonald, S. W. S., & Dixon, R. A. (2002). Variabil ity in reaction time performance of younger and older adults. Journal of Gerontology 57, 101-115. Hultsch, D. F., MacDonald, S. W. S., Hunter, M. A., Levy-Bencheton, J. & Strauss, E. (2000). Intraindividual variability in cognitive perfor mance in older adults: Comparison of adults with mild dementia, adults with arthritis, and healthy adults. Neuropsychology 14, 588598. Katz D. A., & McHorney C. A. (2002). The re lationship between insomnia and health-related quality of life in patients with chronic illness. Journal of Family Practice 51, 229-235. Kreft, I. G. G., de Leeuw, J., & Aiken, L. S. (1995). The effect of different forms of centering in hierarchical linear models. Multivariate Behavioral Research 30, 1-21.

PAGE 88

88 Krueger, G. P. (1989). Sustained work, fatigue, sleep loss and performan ce: a review of the issues. Work Stress 3, 129-141. Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). New York: Oxford University Press. Lezak, M. D. (2004). Neuropsychological assessment (4th ed.). New York: Oxford University Press. Li, S., Aggen, S. H., Nessleroade, J. R., & Baltes, P. B. (2001). Short-term fluctuations in elderly peoples sensorimotor functioning predict te xt and spatial memory performance: The MacArthur Successful Aging Studies. Gerontology 47, 100-116. Lichstein, K. L., Riedel, B. W., & Means, M. K. (1999). Psychological treatment of late-life insomnia. In R. Schulz, G. Maddox & M. P. Lawton (Eds.), Annual review of gerontology and geriatrics: Vol. 18. Focus on inte rventions research with older adults (pp. 74-110). New York: Springer. Lindenburger, U., & Baltes, P. B. (1994). Sensor y functioning and intelligence in old age: A strong connection. Psychology and Aging 9, 339-355. MacDonald, S. W. S., Hultsch, D. F., & Dixon, R. A. (2003). Performance variability is related to change in cognition: Evidence fro m the Victoria Longitudinal Study. Psychology and Aging 18, 510-523. McCoy, K. J. M. (2004). Understanding the tran sition from normal cognitive aging to mild cognitive impairment: Comparing the intraind ividual variability in cognitive function. Dissertation Abstracts International University of Florida. McCrae, C. S., Rowe, M. A., Tierney, C. G., Da utovich, N. D., DeFinis, A. L., & McNamara, J. P. H. (2005). Sleep complaints, subjectiv e and objective sleep patterns, health, psychological adjustment, and daytime func tioning in community-d welling older adults. Journal of Gerontology 60, 182-189. McCrea, C. S., Wilson, N. M., Lichstein, K. L., Durrence, H. H., Taylor, D. J., Bush, A. J., & Riedel, B. W. (2003). Young old and old ol d poor sleepers with and without insomnia complaints. Journal of Psychosomatic Research 54, 11-19. Mini Mitter Co., I. (2001). Actiwatch 16/Actiwatch 64/Actiwatch -L/Actiwatch-Score instruction manual : Mini Mitter Co., Inc. Monk, T. H., Thompson, W. K., Buysee, D. J., Hall, M., Nofzinger, E. A., & Reynolds, C. F. (2006). Sleep in healthy senior s: A diary study of the rela tion between bedtime and the amount of sleep obtained. Journal of Sleep Research 15, 256-260. Morgan, K. (1996). Mental health factors in late -life insomnia. Reviews in Clinical Gerontology 6, 75-83.

PAGE 89

89 Morgan, K. (2000). Sleep and Aging. In K. Lichstein & C. Morin (Eds.), Treatment of late-life insomnia. Thousand Oaks, CA: Sage Publications. Muzur, A., Pace-Schott, E. F., & Hobson, J. A. (2002). The prefront al cortex in sleep. Trends in Cognitive Sciences 6, 475-481. Nesselroade, J. R., & Salthouse, T. A. (2004). Methodological and theoretical implications of intraindividual variability in perceptual-motor performance. Journal of Gerontology 59, 49-55. Neubauer, D. N. (2004). Chronic insomnia: current issues. Clinical Cornerstone 6, S17-S22. Neupert, S. D., Almeida, D. M., Mroczek, D. K., & Spiro, A. (2006). Daily stressors and memory failures in a naturalistic setting: Findings from the VA Normative Aging Study. Psychology and Aging 21, 424-429. Newman, A. B., Enright, P. L., Manolio, T. A ., Haponik, E. F., & Wahl, P. W. (1997). Sleep disturbance, psychological corr elates, and cardiovascular dise ase in 5201 older adults: The cardiovascular health study. Journal of the American Geriatrics Society 45, 1-7. Ohayon, M. (1996). Epidemiological study on insomnia in the general population. Sleep 19, S7S15. Ohayon, M. (2002). Methodology of a study on insomia in the general population. Encphale, 28, 217-226. Olofsen, E., Dinges, D. F., & Van Dongen, H. P. A. (2004). Nonlinear mixed-effects modeling: Individualization and prediction. Aviation, Space, and Environmental Medicine 75, 134140. Pallesen, S., Nordhus, I. H. and Kvale, G. (1998). Nonpharmacological interventions for insomnia in older adults: A meta -analysis of treatment efficacy. Psychotherapy, 35, 472482. Pilcher, J. J., & Huffcut, A. J. (1996). Effect s of sleep deprivation on performance: A metaanalysis, Sleep 19, 318-326. Prinz, P. N., Peskind, E. R., Vitaliano, P. P., Raskind, M. A., Eisdorfer, C., Zemcuznikov, N., & Gerber, C. J. (1982). Changes in the sleep and waking EEGs of nondemented and demented elderly subjects. Journal of the American Geriatric Society 30, 86-93. Roth T., & Roehrs T. (2003). Insomnia: epidem iology, characteristics, and consequences. Clinical Cornerstone 5, 5-15. Salthouse, T. A. (1991). Theoretical perspectives on cognitive aging Hillsdale, NJ: Erlbaum. Salthouse, T. A., & Berish, D. E. (2005). Co rrelates of within-per son (across-occasion) variability in reaction time. Neuropsychology 19, 77-87.

PAGE 90

90 Salthouse, T. A., (1996). The processing-speed theo ry of adult age differences in cognition. Psychological Review 103, 403-428. Salthouse, T. A., Nesselroade, J. R., & Berish, D. E. (2006). Short-term variability in cognitive performance and the calibration of longitudinal change. Journal of Gerontology 61, 144151. Schneider, C., Fulda, S., & Schulz, H. ( 2004). Daytime variation in performance and tiredness/sleepiness ratings in patients with insomnia, narcolepsy, sleep apnea and normal controls. Journal of Sleep Research 13, 373-383. Schwartz S., Anderson, W. M., Cole, S. R., Cornoni-Huntley, J., Hays, J. C., & Blazer, D. (1999). Insomnia and heart disease: a review of epidemiologic studies. Journal of Psychosomatic Research, 47, 313-333. Siegler, R. S. (1994). Cognitive variability: A key to understanding cognitive development. Current Directions in Psychological Science 3, 1-5. Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis : modeling change and event occurrence Oxford ; New York: Oxford University Press. Singer, J. D., Davidson, S. M., Graham, S., & Davidson, H. S. (1998). Physician retention in community and migrant health cente rs: who stays and for how long? Medical Care, 36 1198-1213. Singer, J. D., Fuller, B., Keiley, M. K., & Wolf, A. (1998). Early child-care select ion: variation by geographic location, maternal charac teristics, and family structure. Developmental Psychology, 34 1129-1144. Sliwinski, M. J., & Buschke, H. (2004). Modeli ng intrindividual cognitive change in aging adults: Results from the Einstein Aging Studies. Aging Neuropsychology and Cognition, 11, 196-211. Sliwinski, M. J., Hofer, S. M., & Hall, C. ( 2003). Correlated and coupled cognitive change in older adults with and wit hout preclinical dementia. Psychology and Aging 18, 672-683. Sliwinski, M. J., Smyth, J. M., Hofer, S. M., & Stawski, R. S., (2006). Intraindividual coupling of daily stress and cognition. Psychology and Aging 21, 545-557. Smith, A. (1982). Symbol Digit Modalities Test Revised : Manual. Los Angeles : Western Psychological Services. Smith, A. (1991) Symbol Digit Modalities Test Los Angeles : Western Psychological Services. Stoller, M. K. (1994). Economic effect of insomnia. Clinical Therapist 16, 873-897. Tabachnick, B. G., & Fidell, L. S. (2007). Using Multivariate Statistics 5th Edition Boston, MA: Pearson Education, Inc.

PAGE 91

91 Thurstone, T. G. (1962). Primary mental ability for Grades 9-12 (Rev. ed.). Chicago, IL: Science Research Associates. Totterdell, P., Reynolds, S., Parkinson, B., & Brin er, R. B. (1994). Associations of sleep with everyday mood, minor symptoms and social interaction experience. Sleep 17, 466-475. Tworoger, S. S, Lee, S., Schernhammer, E. S., & Grodstein, F. (2006). Th e association of selfreported sleep duration, difficu lty sleeping, and snoring with cognitive function in older women. Alzheimer Disease and Associated Disorders 20, 41-48. U.S. Census Bureau. (2004) Estimates of Average Annual Rates of the Components of Population Change for the United States and States: April 1, 2000 to July 1, 2004 (NSTEST2004-06) Vallieres, A., Ivers, H., Bastien, C. H., B eaulieu-Bonneau, S. and Morin, C. M. (2005). Variability and predictability in sleep patterns of chronic insomniacs. Journal of Sleep Research, 14, 447-453. Van Dongen, H. P. A., Maislin, G., & Dinges, D. F. (2004). Dealing with inter-individual differences in the temporal dynamics of fatigue and performance: Importance and techniques. Aviation, Space, and Environmental Medicine 75, 147-154. Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F., (2003). The cumulative cost of additional wakefulness: does-respons e effects on neurobehavioral functions and sleep physiology from chronic sleep rest riction and total sl eep deprivation. Sleep 26, 117126. Vignola, A., Lamoureax, C., Bastien, C. H., & Mori n, C. M. (2000). Effects of chronic insomnia and use of benzodiazepines on daytime performance in older adults. Journal of Gerontology 55, 54-62. Vorona, R. D., Winn, M. P., Babino, T. W., Eng, B. P., Fledman, H. R., & Ware, J. C. (2005). Overweight and obese patients in a primary care population repor t less sleep than patients with a normal body mass index. Archives of Internal Medicine 165, 25-30. Wagner, U., Gais, S., & Born, J. (2001). Emoti onal memory formation is enhanced across sleep intervals with high amounts of rapid eye movement sleep. Learning and Memory 8, 112119. Wagner, u., Gais, S., Haider, H., Verleger, R., & Born, J. (2004). Sleep inspires insight. Nature 427, 352-355. Walker, M P., & Stickgold, R. (2004). Sleep-dep endent learning and memory consolidation. Neuron 44, 121-133. Walsh, J. K., & Engelhardt, C. L. (1999). Prev alence and health consequences of insomnia. Sleep, 22, S386-S393.

PAGE 92

92 Webb, W. B. (1982). Sleep in older persons: Slee p structures in 50to 60-year-old men and women. Journal of Gerontology 52, 368-371. Webb, W. B. (1985). A further analysis of age and sleep deprivation effects. Psychophysiology 22, 156-161. Webb, W. B., & Levy, C. M. (1982). Age, sleep deprivation, and performance. Psychophysiology 19, 272-276. Wilkinson, R. T. (1992). The measurement of sleepiness. In Broughton Ogilvie (Eds.), Sleep, arousal, and performance Birkhauser: Sage Publications Yesavage, J. A. (1983). Developm ent and validation of a Geriatri c Depression Screening Scale: A preliminary report. Journal of Psychiatric Research, 17 37-49. Yoo, S. S., Hu, P. T., Gujar, N., Jolesz, F. A., & Walker, M. P. (2007). A de ficit in the ability to form new human memories without sleep. Nature Neuroscience February 2007, 1-8. Zepelin, H., McDonald, C. S., & Zammit, G. K. (1984). Effects of age on auditory awakening thresholds. Journal of Gerontology 39, 294-300.

PAGE 93

93 BIOGRAPHICAL SKETCH My scholarly interests in the aging proce ss began when I was an undergraduate at the University of Nevada, Las Vegas. Through w itnessing the aging of my grandmother and a seminar on cognitive aging a spark was lit in me th at has never gone out. Over the course of my junior and senior year I was aw arded grants from the National In stitute on Health (NIH) and the National Science Foundation (NSF) to conduct independent research. After graduating magna cum laude and earning a Bachelor of Arts degree in psychology I quickly a pplied to graduate school. I was accepted into the Department of Clinical and Health Psychology doctoral program at the University of Florida and awarded a co mpetitive spot on a National Institute on Aging Training Grant (T32). Last year I was honored to be the recipient of the American Psychological Associations Division 20 (Adult Development a nd Aging) Research and Retirement Award for most outstanding proposed masters thesis for the precursor to this document and was recently awarded a Most Outstanding Poster award by th e college of Public Health and Health Professionals for presentation of sections of th is thesis. My research interests are many, but always share the common theme of aging. I am fa scinated by the cognitive aging process, sleeps relation to cognitive performance, and short-term variability.


Permanent Link: http://ufdc.ufl.edu/UFE0020955/00001

Material Information

Title: Intraindividual Variability in Objective and Subjective Sleep and Cognitive Performance in Older Adults with Insomnia
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0020955:00001

Permanent Link: http://ufdc.ufl.edu/UFE0020955/00001

Material Information

Title: Intraindividual Variability in Objective and Subjective Sleep and Cognitive Performance in Older Adults with Insomnia
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0020955:00001


This item has the following downloads:


Full Text






INTRAINDIVIDUAL VARIABILITY INT OBJECTIVE AND SUBJECTIVE SLEEP AND
COGNITIVE PERFORMANCE INT OLDER ADULTS WITH INSOMNIA


















By

JOSEPH M. DZIERZEWSKI


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2007































O 2007 Joseph M. Dzierzewski



































To my Mom and Dad... for everything.









ACKNOWLEDGMENTS

There are so many people that have contributed to this proj ect in one way or another. First,

and foremost, I must thank my chair, Dr. Christina McCrae, for her mentorship. She has given

me the support I needed to grow as a scientist. I must also thank my co-chair, Dr. Michael

Marsiske, for his encouragement and guidance. I have been truly fortunate to have two mentors

willing to take part in my educational experience. I also express my gratitude to the team of

researchers in the Sleep Research Lab: Amanda Ross, Joseph MacNamara, Natalie Dautovich,

Ashley Stripling, and countless undergraduate research assistants.

At this point I have to turn my appreciation to my parents, Steve and Karen Dzierzewski,

for never holding back. They were always there, always caring, and always believing in me.

They have given me everything and asked for nothing but my happiness in return. I couldn't

have asked for better parents. I owe them everything.

At this point I have to turn my appreciation to my parents, Steve and Karen Dzierzewski,

for never holding back. They were always there, always caring, and always believing in me.

They have given me everything and asked for nothing but my happiness in return. I couldn't

have asked for better parents. I owe them everything.












TABLE OF CONTENTS


page

ACKNOWLEDGMENTS .............. ...............4.....


LIST OF TABLES ................ ...............8............ ....


LIST OF FIGURES .............. ...............9.....


AB S TRAC T ............._. .......... ..............._ 10...


CHAPTER


1 INTRODUCTION ................. ...............12.......... ......


2 REVIEW OF THE LITERATURE ................. ...............14.......... ....


Older Adults ................. ...............14.................

Sleep in Older Adults ................. ...............14................
"Normal" Sleep .............. ...............14....
Insomnia ................ ...............15.................

Sleep and Cognition.................. ...................1
Sleep and Cognition in Older Adults............... ...............20.
Intraindividual Variability (IIV) ................ ...............21........... ....
Variability in Sleep ................. ...............22........... ....
Variability in Cognition............... ...............2
Summary ................. ...............26.................

3 STATEMENT OF THE PROBLEM ................. ...............28........... ...


Aim 1: To Determine the Amount of Variability in Sleep and Cognition Found Within
Older Insomniacs Compared to the Amount of Variability Between Older Adults with
Insomnia ................. ...............29.................

Importance ................. ...............29.................
Hypothesis ................ ...............29.................
A analysis .................. ... .... ......... ....... .... .... ... ... ........2
Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive
Measures in Older Adults with Insomnia. ............. ...............30.....

Importance ................. ...............30.................
Hypothesi s ................. ...............30._._._......
A naly si s .........._..... ........._ .. ...._._ .. .. .. .... .... ........
Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-to-
Day-Level) Association between Sleep and Cognition in Older Adults with Insomnia.....31
Im portance ................. ...............3.. 1..............
Hypothesi s ................. ...............3.. 1..............
Analy si s ................. ...............3.. 1..............











4 M ETHODS .............. ...............36....


Main Study............... ...............36.
Participants .............. ...............36....
Inclusion Criteria ................. ...............36......... ......
Exclusion Criteria............... ...............37
S ample Characteristics .............. ...............37....
M measures ................. ...............38......... ......
Sl eep Measures ....._.. ................. .................3 8....
Subj ective sleep measures ....._.._................. ........__. ........ 3
Obj ective sleep measures ................. ...............39........ .....
Cognitive M measures ................. ...............40........ ......
Reasoning ................. ...............40........ ......
Processing speed .............. ...............41....
Proc edure s .................. ......_.._ ...............41......
Baseline Study Procedure ......_. ................. ........_.._.........4
Alternate Forms of the Cognitive Measures ......_.._............... ............... 42. ...
Missing Data............... ...............43..

5 RE SULT S .............. ...............46....


Aim 1: To Determine the Amount of Variability in Sleep and Cognition Found Within
Older Adults with Insomnia Compared to the Amount of Variability Between Older
Adults with Insomnia ........._....... ..... ....... ........ _._. ... .. .... ........4
Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive
Measures in Older Adults with Insomnia .............. ...............47....
Associations of Within-Person Variability ......_.._............... ............... 47. ....
Associations of Mean-Level Performance ................. ......._.._ .... ........ ....... ........4
Associations between Within-Person Variability and Mean-Level Performance ...........48
Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-to-
Day-Level) Association between Sleep and Cognition in Older Adults with Insomnia.....49
M ulticollinearity .............. .... ........... ...............49.......
Within-person multicollinearity .............. ...............50....
Between-person multicollinearity .............. ...............50....
Multilevel Model for Letter Series ................ ...............50........... ...
Multilevel Model for Symbol Digit............... ...............51.

6 DI SCUS SSION ................. ...............64................


Review of Findings ............... ... .... ........ ................ ... .. .. .. .. .. ........6
Aim 1: To Determine the Amount of Variability in Sleep and Cognition Found
Within Older Adults with Insomnia Compared to the Amount of Variability
Between Older Adults with Insomnia. ................... ... ......... .. .......... ...........6
Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive
Measures in Older Adults with Insomnia .............. ...............64....











Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-
to-Day-Level) Association between Sleep and Cognition in Older Adults with
Insomnia................ ...............6
MLM for reasoning .............. ...............66....
MLM for processing speed .............. ...............67....
Study Limitations............. .. .................6
Theoretical and Empirical Implications. ........._... ...... .___ .....____ ...........6
Future Directions .............. ...............72....


APPENDIX


A SLEEP DIARY ........._... ...... ..... ...............73....

B LETTER SERIES TEST............... ...............74..


C SYMBOL DIGIT TEST .............. ...............76....

D CORRELATIONS OF PREDICTOR VARIABLES NOT CONTROLLING FOR TIME...78

E REASONING MODELS AT EACH STEP OF MLM BUILDING ............... ... ........._._._.79


F PROCESSING SPEED MODELS AT EACH STEP OF MLM BUILDING........................82


LIST OF REFERENCES .....___................. ........____.........8

BIOGRAPHICAL SKETCH .............. ...............93....











LIST OF TABLES


Table page

3-1 Level 1 and Level 2 Equations at each Step of the MLM Building Process ..................35

4-1 Participant Characteristics .............. ...............45....

5-1 Amount of within and between-person variability. ...........__.....___ .............. .53

5-2 Correlations of within-person standard deviations. ..........._ ..... .__ ........... ....55

5-3 Correlations of sample means. .............. ...............56....

5-4 Correlations of within-person standard deviations and sample means. ...........................57

5-5 Correlations of within-person predictors. ................ ......... ......... ...........5

5-6 Correlations of between-person predictors. ............. ...............59.....

5-7 Steps taken in building the Letter Series Multilevel Model. ............. .....................6

5-8 Sleep variables predicting reasoning. ............. ...............61.....

5-9 Steps taken in building the Symbol Digit Multilevel Model. ............. ......................6

5-1 Sleep variables predicting processing speed ................. ...............63...............

A-1 Example of a Sleep Diary .............. ...............73....

D-1 Correlations of within-person, with time, predictors ................. ............................78

D-2 Correlations of between-person, with time, predictors ................. .......... ...............78

E-1 Reasoning MLM Step 2: Adding Time. ............. ...............79.....

E-2 Reasoning MLM Step 3: Adding Sleep. ............. ...............80.....

E-3 Reasoning MLM Step 4: Adding Interactions. .............. ...............81....

F-1 Processing Speed MLM Step 2: Adding Time. ............. ...............82.....

F-2 Processing Speed MLM Step 3: Adding Sleep. .............. ...............83....

F-3 Processing Speed MLM Step 4: Adding Interactions ................. ......____........._....84










LIST OF FIGURES

FiMr IM Le

4-1 Overall design of REST Study ................. ...............44.._.._ ....

5-1 Relative amount of within-person variability ........._._.__........ ......_.. .........5









Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

INTRAINDIVIDUAL VARIABILITY INT OBJECTIVE AND SUBJECTIVE SLEEP AND
COGNITIVE PERFORMANCE INT OLDER ADULTS WITH INSOMNIA

Joseph M. Dzierzewski

May 2007

Chair: Christina McCrae
Cochair: Michael Marsiske
Major: Psychology

Sleep researchers and theorists alike have speculated for many years that one of the

primary purposes of sleep is to provide a period of rest for the brain that is essential for

maintaining optimal cognitive functioning. Although many have attempted to systematically

examine this relationship, the literature is plagued by inconsistent and contradictory findings.

The current study examined the sleep-cognition relationship in a sample of community-dwelling

older adults with insomnia.

Forty-eight older adults with insomnia (Mean Age = 69.91 years, SD = 7.24) concurrently

wore wrist actigraphy while completing sleep diaries and a measure of perceptual speed and

reasoning for fourteen consecutive days. Descriptive analysis revealed that individuals exhibited

a substantial amount, between 48% and 189%, of variability in sleep and cognition within-person

as compared to between-person. Correlational analysis revealed that more variable sleep is

associated with poorer reasoning and better perceptual speed.

Multilevel modeling (MLM) indicated that days with longer than average sleep onset are

associated with better than average reasoning, as is spending, on average, greater time resting in

the morning. However, days when an individual, who usually spends long periods to fall asleep,

takes even longer to fall asleep are associated with poorer reasoning. Spending above-average









amount of time awake in the middle of the night is associated with below-average perceptual

speed. However, days when an individual, who usually spends long periods resting in the

morning, rests for even longer in the morning are associated with better processing speed.

Results lend support to the sleep-cognition relationship. These findings illustrate the utility

of studying intraindividual variability in sleep. There is a daily relationship between sleep, or

better stated wake, parameters and cognitive performance in older adults with insomnia.









CHAPTER 1
INTTRODUCTION

The current study seeks to examine the intraindividual variability (i.e., within-person

variability) of obj ectively and subj ectively measured sleep and cognitive performance in older

adults with insomnia. Specifically, this study will address questions concerning the amount of

within-person variability in sleep and cognition, how this variability might be related across the

two domains and how it is related to average performance/behavior, and how changes in daily

sleep might relate to changes in daily cognitive performance.

There is a large amount of research illustrating that normal aging is associated with

decreases in sleep and decreases in cognitive ability. Given the known decreases in sleep and

cognitive performance associated with increased age, it is logical to think that these two

phenomena might be related. In fact, many studies have proven this to be an accurate assumption

by means of sleep depriving individuals and then testing their cognitive capabilities. However,

the ecological validity of such studies is low because individuals do not tend to actively keep

themselves awake for such extended periods of time. And, though generally decrements in

functioning are observed following deprivation, some studies have found no such results. An

interesting interpretation of these mixed results is that people vary in their susceptibility to sleep

loss (i.e., there are interindividual differences in intraindividual processes).

A current line of research, and analysis, within the developmental and cognitive aging

fields is the study of intraindividual variability in cognitive functioning. Results from this context

have indicated that within-person variability is both a normal homeostatic "hum" and large

inconsistency can be indicative of impaired neuronal functioning and that variability tends to

increase with advancing age. Furthermore, statistical advances associated with the study of

intraindividual variability allows for the testing of between-person, within-person









differences/associations, and between-person differences on within-person processes. If

individuals with impaired neural functioning exhibit high within-person variability and sleep loss

is expected to cause decreased cognitive performance through its effects on the prefrontal cortex

then the application of intraindividual variability methodology to the study of the sleep-cognition

relationship in older insomniacs is warranted and should prove fruitful.

The current study should expand the knowledge domain in the fields of sleep research and

cognitive aging while bridging two fields that are very complimentary yet consistently distant

from one another. It is hypothesized that sleep and cognition will both display considerable

amounts of within-person variability, that variability will be related across constructs, and that

daily changes in sleep will coincide with daily changes in cognitive performance.

These hypotheses have implications for our current understanding of the purpose of sleep,

of the effects of sleep loss, and causes of age-related cognitive decline. Possible implications

could be sleep-based treatments to deal with one possible source of age associated cognitive

decline, the need to assess sleep when conducting cognitive assessment, and the need to assess

cognition multiple times to get an accurate indication of true level.

The following chapters provide a comprehensive overview of the relevant sleep and

cognitive literature. This will be followed by the specific aims, hypotheses, and analysis of each

part of the study. Results will then be presented and discussed.










CHAPTER 2
REVIEW OF THE LITERATURE

Older Adults

There are currently over 36 million adults living in the United States over the age of 65.

This number is expected to exceed 64 million in the next 40 years and it has been estimated that

the number of individuals between the age of 65 and 85 will grow by approximately 1 13%

between 2000 and 2050 while the number of individuals over 85 years old will experience 388%

growth in the same 50 years (U.S. Census Bureau, 2004). With this rapidly increasing population

it is essential for psychologists to expand our knowledge of the various factors affecting the

agmng process.

Sleep in Older Adults

"Normal" Sleep

The patterns, durations, frequencies, and correlates of sleep in older adults are all well

studied phenomenon. In general, as an individual increases in age their sleep becomes lighter,

shorter and more fragmented than when they were younger (Morgan, 2000). It has been found

that older adults experience more frequent shifts from one sleep stage to another, more frequent

intrasleep arousals (Bosselli, Parrino, Smerieri, & Terzano, 1998), and more, and longer, periods

of alpha activity during sleep (Webb, 1982) than do their younger counterparts.

There is currently considerable debate about whether or not rapid-eye-movement (REM)

sleep decreases with age (Bliwise, 1993). However, it is generally agreed upon that one of the

most prominent changes in sleep architecture associated with the aging process is the steady and

drastic decrease in the amount of time spent in deep, slow-wave-sleep (Stages 3 and 4) (SWS;

Prinz et al., 1982). The decrease in the amount of time spent in SWS in late-life means that older

adults spend much more time in the light, non-restorative sleep Stages 1 and 2. These changes









have led to the sleep of older adults being characterized as "structurally lighter" than that of

younger adults. In fact, it has been found that older adults awaken more easily from sleep than do

younger adults (Zepelin, McDonald, & Zammit, 1984).

Whether or not these age related changes in sleep are part of the "normal" aging process or

if they are the behavioral manifestations of an underlying pathology is debatable. A recent meta-

analysis of the age-related sleep changes confirm that increased age is associated with increased

wake time after sleep onset and sleep onset latency and decreased total sleep time at medium to

high levels (Floyd, Medler, Ager, & Janisse, 2000). In general, the functions of sleep have not

yet been agreed upon but it is the consensus of the maj ority of sleep researchers that the above

described changes in sleep reflect "normal ontogenetic change" (Morgan, 2000).

Insomnia

Insomnia, on the other hand, is among the most prevalent disorders of late-life. The

Diagnostic and Statistical Manual of2~ental Disorders 4' Edition (DSM-IV) defines insomnia

as the difficulty initiating or maintaining sleep, or non-restorative sleep, for at least one month

that causes significant distress (American Psychiatric Association, 1994). The prevalence of

insomnia in older adults is not agreed upon. However, it is known that the prevalence does

increase with increasing age (Ohayon, 1996). In an epidemiological community-based sample of

over 5,000 older adults estimates of the prevalence of insomnia were 65% (Newman, Enright,

Manolio, Haponik, & Wahl, 1997) indicating the widespread nature of the disorder. Furthermore,

insomnia in older adults tends to be a chronic condition with the average span of the disorder

lasting 12 years (McCrae et al., 2003).

Insomnia in late-life is not a solitary event; it is accompanied by many unwanted

consequences and correlates. Individuals with insomnia have been shown to experience quality

of life hindrances equivalent to the experience of congestive heart failure patients (Katz &









McHorney, 2002). Other studies have reported deficits in cognitive and psychomotor

functioning, including memory, concentration, attention, reasoning, problem solving, and

reaction time (Harrison & Horne, 2000; Roth & Roehrs, 2003).

Additionally, insomnia has serious negative effects on mental health and social

functioning. Insomnia is often co-morbid with psychiatric, mood and anxiety disorders.

Estimates suggest that 30% to 50% of individuals with insomnia also have an accompanying

psychiatric disorder (Benca, 2001; Morgan, 1996). Insomnia has been associated with the

occurrence of coronary heart disease (Schwartz et al., 1999) and obesity (Vorona et al., 2005).

The negative impact of insomnia does not stop at significant personal hardship. Insomnia

has significant national economic consequences. Insomnia has been associated with a significant

increase in absenteeism due to health problems (twice as likely to miss work), lost productivity at

work, increased rates of health care utilization (after controlling for age, sex, and medical and

psychiatric disease) and automobile accidents (insomnia increases the risk of traffic accidents,

accidents at home, and public accidents by 200% to 300%, and work-related accidents by 150%)

(Benca, 2001; Hublin & Partinen, 2002; Katz & McHorney, 2002; Neubauer, 2004; Roth &

Roehrs, 2003;). The total economic burden of insomnia has been estimated to be between a

staggering $77 and $100 billion per year (Hublin & Partinen, 2002; Stoller, 1994; Walsh &

Engelhardt, 1999).

Sleep and Cognition

The link between sleep and cognition has long been theorized and studied. Although this

relationship seems rather intuitive there is a lack of consensus regarding the effects of sleep loss

on cognitive functions. For reasons of experimental control most studies of the sleep-cognition

relationship use a sleep deprivation paradigm. In this paradigm, participants are usually










cognitively tested prior to deprivation, kept awake in a laboratory for an extended period of time,

and then tested again.

In a recent review of the sleep deprivation cognition literature, Harrison and Horne

(2000) summarize the extant mixed findings regarding the cognitive consequences of laboratory

induced sleep loss as being "unclear from the literature whether tasks associated with cognitive

speed, psychomotor skills, auditory and visual attention, and short-term attention are sensitive to

1 night of sleep deprivation for any other reason than their monotony and lack of novelty". This

view is shared by scientists who argue the sleep loss has no specific physiological or behavioral

manifestations but operates through a global reduction in arousal (Wilkinson, 1992). Support for

this view has been drawn from studies illustrating difficult and important assessments, such as

IQ test, are resilient to 36 hours or more of sleep deprivation (Horne, 1988).

There are, however, a growing number of neuropsychological and imaging studies that

have illustrated the prefrontal region of the cerebral cortex is particularly sensitive to sleep loss

and that the executive functions (i.e., working memory, attention, and processing speed) are

dependent on processes that occur in this brain region, suggesting that these functions are the

most susceptible to sleep loss (Drummond et al., 2000). Yet, many sleep researchers adhere to

the stance that divergent skills, as opposed to convergent skills, are the ones most affected by

sleep loss (Harrison & Horne, 2000). And early reviewers of the sleep deprivation cognition

relationship suggested a negative impact of sleep loss on reaction times and vigilance (Krueger,

1989).

Recent evidence from more physiologically based research suggests sleep provides an

exclusive form of recovery for the cerebral cortex, especially the prefrontal cortex (Horne, 1993).

The prefrontal cortex, which has been shown to be vital for optimal cognitive functioning, has









been shown to recover from strains placed on it during wakefulness during sleep. This can be

seen in its high voltage and slowest brainwaves, compared to other brain regions, while in

NREM sleep (Muzur, Pace-Schott, & Hobson, 2002). The prefrontal cortex is thought to be the

hardest working region of the brain and it is therefore believed that it is also the region of the

brain that needs the most recovery during sleep. Hence, the prefrontal cortex and its associated

functions are believed to be the most susceptible to the loss of sleep (Durmer & Dinges, 2005).

In fact, marked changes in activation of the prefrontal cortex have been observed following 30-

35 hours of sleep deprivation (Drummond, 2000), as were detriments in performance on a known

and agreed upon task of executive functioning, the Tower of London Test (Horne, 1988).

In a related, but different, approach Pilcher and Huffcutt (1996) performed a meta-analysis

on the effects of sleep deprivation on cognitive performance. The results of their analysis, which

included a combined sample of over 1,900, pointed to the intriguing finding that not all sleep

deprivation is created equal. They found that partial sleep deprivation, getting less than 5 hours

of sleep a night, resulted in the largest decrease in cognitive performance; more so than did

short-term total sleep deprivation, less than 45 hours of sleep deprivation, and long-term total

sleep deprivation, more than 45 hours of sleep loss. The effect size for partial sleep deprivation

and cognitive performance was considerably large (d = -3.01). Interestingly, the authors also

reported large effect sizes for the relationship between partial sleep deprivation and simple and

complex cognitive tasks (Pilcher & Huffcutt, 1996).

Recent researches now suggest that total sleep deprivation has a more profound effect on

cognitive functioning than partial sleep deprivation (Van Dongen, Maislin, Mullington, &

Dinges, 2003). With regard to chronic partial sleep deprivation, which most realistically mimics

'real' sleep, the effects on cognitive performance have been mixed. Repeated days of sleep









restriction to between 3 and 6 hours of sleep per night have been experimentally shown to

negatively affect cognitive speed, working memory, and attention (Belenky et al., 2003; Dinges

et al., 1997; Drake et al., 2001; Van Dongen et al., 2003). Interestingly, it has been shown that

the cumulative effect of 14 days of sleep restriction to 4 hours per night was cognitively

(attention, working memory, cognitive throughput) equivalent to 2 nights of total sleep

deprivation and 14 days of restriction to 6 hours per night was equivalent to 1 night of total sleep

deprivation (Durmer & Dinges, 2005). The cognitive deficits observed in chronic sleep

restriction do not accumulate in an additive fashion (i.e., a total of 20 hours awake in sleep

restriction over 4 days does not equal 20 hours straight of sleep deprivation). This has been

interpreted as adaptation in sleep restriction (Drake et al., 2001) where individuals leamn to deal

with the effects of sleep loss.

There is a growing amount of evidence that points to the role of sleep for memory

consolidation. REM sleep is suspected to influence procedural learning and emotional memories

(Gais & Bomn, 2004; Wagner, Gais, & Bomn, 2001) while NREM sleep has been implicated in

declarative memory processes. Sleep may also facilitate the realization of underlying rules or

insight gaining (Wagner, Gais, Haider, Verleger, & Born, 2004). Sleep has been shown to be

essential for memory consolidation after learning has occurred (Walker & Stickgold, 2004) and

more recently the need for sleep prior to the acquisition of information has been demonstrated.

Subj ects who experienced 3 5 hours of sleep deprivation prior to exposure to new information

had significantly worse recall of that information than did normal control sleepers (Yoo, Hu,

Guj ar, Jolesz, & Walker, 2007). Sleep appears to be important for the consolidation of previously

learned information and for the acquisition of new information. See Hobson & Pace-Schott

(2002) for a review of the neural systems associated with sleep and learning.









Both the authors of the review (Harrison & Horne, 2000) and the meta-analysis (Pilcher &

Huffcutt, 1996) point to the need for more research on the sleep and cognition relationship in

older adults.

Sleep and Cognition in Older Adults

The two studies that have examined the effects of sleep deprivation on cognitive

performance in older adults found worse cognitive performance following sleep deprivation in

older adults than in younger adults (Webb, 1985; Web & Levy, 1982). In a sample of over 2,500

older women with osteoporosis poor obj ective sleep was found to be associated with impaired

overall cognitive functioning, measured by the MMSE, and decreased processing speed,

measured by time to complete Trails B (Blackwell et al., 2006). Similarly, in a sample of nearly

2,000 older community-dwelling women, those who sleep less and had more difficulty initiating

and maintaining sleep performed cognitively worse than those whose sleep could be

characterized as 'good' (Tworoger, Lee, Schernhammer, & Grodstein, 2006). In a sample of over

6,000 older men, insomnia symptoms were found to be independent predictors of three year

cognitive decline independent of demographic, behavioral, and health factors (Cricco,

Simonsick, & Foley, 2001).

Although much research has been conducted on the relationship between sleep and

cognitive performance, the same can not be said for studying this relationship in older adults

with insomnia. The few studies that have been completed on this sample indicate that there is a

sleep cognition relationship in older adults with insomnia. For example, it has been found that

older adults with insomnia have a deficiency in slow-wave-sleep that is related to processing

speed but that this same slow-wave-sleep processing speed relationship is not present in normal

sleeping older adults (Crenshaw & Edinger, 1999). In a study comparing good sleeping older

adults to medicated and un-medicated older insomniacs on the relationship between obj ective









and subj ective sleep and cognitive performance found that older insomniacs on and off

benzodiazepines performed worse on measures of attention and concentration (Vignola,

Lamoureax, Bastien, & Morin, 2000). In a follow-up study, it was found that the direction and

strength of the various sleep parameters relationship to cognitive performance differed by group.

In general, as sleep worsened in the good sleepers, cognitive performance decreased. However,

mixed findings were reported for both medicated and un-medicated insomniacs. As some sleep

parameters increased so did cognitive performance while other relationships were unexpectedly

found in the opposite direction (e.g., as subj ective total wake time increased so did memory)

(Bastein et al., 2003).

Intraindividual Variability (IIV)

For behavioral scientists to fully and effectively represent processes and change within

individuals over time, the employment of intraindivdual variability methodology has been

termed "absolutely essential" (Nesselroade, 2002). Intraindividual variability (IIV) is any change

in performance over a short period of time. In this sense, it can be viewed as an individual's

fluctuation around their mean. Intraindividual variability can be conceptualized as an

individual's mean standard deviation; that is, their average inconsistency or consistency around

their usual performance.

The study of inraindividual variability, or within-person variability, is vitally important for

several reasons. First, it allows variance components to be decomposed into their respective

within-person and between-person parts (Kreft, de Leeuw, & Aiken, 1995). This is essential

because only when it is know how much variability is located within and between individuals

can one attempt to explain that variability. Another key reason to study intraindividual variability

is in the interpretation of the results. The ecological fallacy refers to applying results obtained at

the group-level to individuals and the atomistic fallacy is applying results obtained within an









individual to groups (Tabachnik & Fidell, 2007). The study of intraindividual variability allows

researchers to comment on processes occurring both within individuals and between groups.

Variability in Sleep

Experts have commented on the relevance of individual variability in sleep patterns for

broadening our understanding of sleep (Espie, 1991; Pallesen, Nordhus, & Kvale, 1998).

Although IIV has received some attention in the sleep literature, the maj ority of research has

focused on characterizing variability in the sleep patterns of good and poor sleepers.

Specifically, researchers have demonstrated that individuals with insomnia tend to exhibit highly

variable sleep patterns (Coates et al., 1981; Edinger et al., 1997; Edinger, Marsh, Mccall, Erwin,

& Lininger, 1991; Frankel, Coursey, Buchbinder, & Snyder, 1976; Hauri & Wisbey, 1992;

Vallieres, Ivers, Bastien, Beaulieu-Bonneau, & Morin, 2005) while normal sleepers tend to

exhibit less variable sleep patterns (Edinger et al., 1997; McCrae et al., 2005; McCrae, Wilson,

& Lichstein, 2003).

In examining the relationship between bedtime and total sleep time (TST) and time in bed

(TIB), it was found the approximately 50% of the total variability in these two variables (TST

and TIB) were found within older healthy sleepers (Monk et al., 2006). However, these

researchers did not attempt to systematically explain within and between-person variability

separately. Recent research examining within-group variability in disordered sleepers

(insomniacs, narcoleptics and individuals with treated and untreated obstructive sleep apnea) has

shown that persons with sleep apnea and narcolepsy have more variable daytime functioning and

poorer cognitive functioning (concentration and attention) than control good sleepers and

insomniacs perform worse cognitively but have more consistent performance across a ten hour

testing session (Schneider, Fulda, & Schulz, 2004). In a 14 day diary study of 30 younger adults,

it was found that variability in "cognitive symptoms" was predicted by sleep latency onset, and










quality (Totterdell, Reynolds, Parkinson, & Briner, 1994). Once again, the prediction over the

course of days was not separated into within and between-person components.

Several studies of the effects of sleep loss on cognitive performance have examined

sources of variance in various neurobehavioral functions (Olofsen, Dinges, & Van Dongen,

2004; Van Dongen, Maislin, & Dinges, 2004). It was found that well over 50% of the variance in

digit-symbol (ICC = .82) and psychomotor vigilance (ICC = .69) was a product of "stable

interindividual differences" in vulnerability to sleep loss (Van Dongen et al, 2004). Neither study

made an attempt to systematically explain the within-person variations observed.

Variability in Cognition

The extent literature examining within-person variability is vast and growing rapidly.

There are, however, many inconsistencies within the literature. For example, Salthouse,

Nesselroade, and Berish (2006) examined the intraindividual variability in thirteen cognitive

tasks in 113 individuals, over three occasions, between the ages of 18 and 97 years old. They

found that within-person variability was approximately 50% the size of between-person

variability estimates, that increased within-person variability was related to decreased

performance, and that within-person variability estimates were not related across constructs

(Salthouse, Nesselroade, & Berish, 2006). However, in a similar study Nesselroade and

Salthouse (2004) examined the within-person variability in perceptual-motor performance in 204

individuals between the ages of 20 and 91 years old, also over three occasions. Findings from

this study indicate that intraindividual variability is substantial (~50% of interindividual

variability), variability is trait-like (i.e., operates consistently within people over time), increased

variability is associated with poorer cognitive functioning, and increased age is associated with

increased within-person variability (Nesselroade & Salthouse, 2004).









Within-person variability in reaction time has been found to be indicative of neurological

dysfunction. Results of a study examining intraindividual variability indicate that adults with

mild dementia show more day-to-day variability than either adults with arthritis or healthy

adults. Furthermore, across groups, individuals who were more variable on one task were more

variable on other tasks and more likely to performance worse than their less variable

counterparts. Strikingly, within-person variability was found to be the single-most important

predictor of group membership (Hultsch, MacDonald, Hunter, Levy-Bencheton, & Strauss,

2000). Hultsch, MacDonald, and Dixon (2002) reported that in an sample of nearly 800 older

adults increased age was associated with increased within-person variability and poorer

performance on several measures of cognitive ability (perceptual speed, working memory,

episodic memory, and crystallized ability). They also report that variability estimates were

unique predictors of cognitive functioning independent of mean-level performance (Hultsch,

MacDonald, & Dixon, 2002).

The maj ority of the studies examining the association between increased age and within-

person variability are cross-sectional. In a six year longitudinal study of cognitive performance in

healthy older adults that measured performance every two years, several findings of interest were

reported. First, increasing age was associated with increased intraindividual variability in

reaction time. Second, within-person inconsistency in reaction time at measure I was predictive

of subsequent cognitive (reasoning, working memory, processing speed, and episodic memory)

decline. Third, variability increased within individuals longitudinally. Fourth, at measurement

occasions when individuals were more variable they also performed worse on the cognitive

measures (MacDonald, Hultsch, & Dixon, 2003).









In the study assessing intraindividual in cognitive performance in older adults over the

greatest number of measurement occasions (120), it was found that within-person variability was

related across cognitive domains (i.e., people are consistently inconsistent) and that increased

variability was associated with increased overall performance in reasoning, memory, and

processing speed (Allaire & Marsiske, 2005). The authors, as well as other, propose that

variability can serve as an adaptive function and does not always have to be an indication of

deterioration (Li, Aggen, Nesselroade, & Baltes, 2001; Siegler, 1994).

In a further advancement of the literature and in an attempt to capture within-person

variability in reaction times in a naturalistic setting, Salthouse and Berish (2005) provided older

adults with hand held computers that randomly prompted reaction time tests over several days.

Results from this study indicate that within-person variability is as large (and sometimes larger)

than between-person variability but that mean-level performance on reaction time tasks are better

predictors of cognitive performance (Salthouse & Berish, 2005). However, in the attempt to

obtain more ecologically valid results, the authors lost much of their experiment control and

noted that many unmeasured confounds could have been operating systematically within their

data set.

Several studies have used intraindividual variability indexes to categorize different groups

of individuals. Variability in affect and physical functioning (e.g., grip strength and finger

taping) has been shown to differentiate between healthy individuals and individuals who

sustained mild and severe head injuries. Individuals with head injuries were found to be more

variable across domains and performed worse on average (Burton, Hultsch, Strauss, & Hunter,

2002). It has been found that within-person variability in cognitive performance (memory, speed,

and fluency) is nearly as large, or larger than the between-person variability in older adults with










preclinical dementia but that the amount of within-person variability is much smaller in

individuals without this disorder (Sliwinski, Hofer, & Hall, 2003). Results indicate variability

may be an indicator of neurological compromise.

Recently, researchers have attempted to capture the covariation, or 'coupling', of variables

within individuals. A significant amount of within-person variability has been found in memory

failures (41%), and daily fluctuations in stressful events have been shown to predict both same

day memory problems and next day memory problems in over 300 older adults over the course

of eight days (Neupert, Almeida, Mroczek, & Spiro, 2006). In a study examining the relationship

between daily stress and daily cognitive performance, it was found that on days when a stressful

event occurred both younger and older adults had worse attention/concentration performances

and were more variable in their cognitive performances (Sliwinski, Smyth, Hofer, & Stawski,

2006). In a related study, it was found that within-person fluctuations in speed of processing

predicted within-person change in memory (Sliwinski & Buschke, 2004).

Summary

The literature suggests the application of studying intraindividual variability in studies of

the sleep-cognition relationship in older adults is warranted. To summarize this review several

six key factors should be noted:

1. The number of older adults is growing rapidly and older adults are faced with unique
concerns that require specialized investigations.

2. Sleep decreases in quality and amount as individuals increase in age.

3. The sleep-cognition relationship has been examined extensively through sleep
deprivation studies. Although results from such studies are not always consistent, it is
generally agreed upon that sleep loss impairs cognitive function especially on tasks that
tap the prefrontal cortex.

4. The study of intraindividual variability is essential for behavior scientists to accurately
portray change and processes within individuals.









5. Insomniacs tend to have highly variable sleep. Sleep researchers have commented that
interindividual differences in susceptibility to sleep loss may be responsible for the
inconclusive findings on the sleep-cognition relationship. Yet, there has been a general
lack of studies of intraindividual variability.

6. Research from the cognitive aging field has shown that within-person variability is large;
usually 50% of the amount of between-person variability, and age is associated with
increased within-person variability. Mixed results have been offered with regard to the
stability of variability and the good-bad distinction of variability.









CHAPTER 3
STATEMENT OF THE PROBLEM

For quite a long time researchers and theorists alike have speculated that one of the main

purposes of sleep is to provide a restorative period for the brain that is essential to maintain

optimal levels of cognitive functioning. In fact, many researchers have attempted to capture this

sleep-cognition relationship (e.g., Harrison & Horne, 2000; Pilcher & Huffcutt, 1996). Given the

normal decrease in amount of time spent asleep (Floyd, Medler, Ager, & Janisse, 2000), the

increased prevalence of insomnia in late-life (Ohayon, 2002), and the normal cognitive decline

associated with increased age (Craik & Byrd, 1982; Hasher & Zacks, 1988; Lindenburger &

Baltes, 1994; Salthouse, 1996), there has been a good deal of research on the sleep-cognition

relationship in older adults (Bastien et al., 2003; Blackwell et al., 2006; Crenshaw & Edinger,

1999; Cricco et al., 2001; Tworoger et al., 2006). The conventional methods employed by the

maj ority of these studies include having participants sleep for one or several nights in a

laboratory and complete a cognitive assessment the following day. Assessed in this manner, the

sleep-cognition relationship has been an elusive one to capture.

Recent research, and advances in methodology and analyses, in the cognitive aging field

into what has been termed "intraindividual variability" (IIV) has produced interesting results

about within-person variation and relationships between cognitive performance and many

covariates (e.g. Nesselroade & Salthouse, 2004; Sliwinski & Buschke, 2004). The study of

intraindividual variability has been catapulted into the forefront of aging research by several

findings indicating that variability indices are better predictors of performance than mean-level

indices (Butler, Hokanson, & Flynn, 1994; Eizenman, Nesselroade, Featherman, & Rowe, 1997).

Recently, prominent sleep researchers have commented on the importance of variability in fully

understanding sleep (Espie, 1991; Pallesen et al., 1998).










The current study applies IIV methodology and analyses to the study of the sleep-cognition

relationship in older adults with insomnia. To this end, the present study addresses the following

three aims:

Aim 1: To Determine the Amount of Variability in Sleep and Cognition Found Within
Older Adults with Insomnia Compared to the Amount of Variability between Older Adults
with Insomnia

Importance

Typical research focuses only on the differences found between people, ignoring the fact

that people themselves are often highly inconsistent. If it is found that a significant amount of

variability is found within-persons, the further examination of intraindividual variability is

warranted.

Hypothesis

Based on prior research it is expected that the variability within-persons will be at least

50% of that found between-persons (Nesselroade, & Salthouse, 2006).

Analysis

To control for any practice effects, or systematic growth in the data, all variables (sleep

and cognitive) will be de-trended prior to calculation of indexes of within and between-person

variability. To de-trend the data, linear regressions will be ran with all sleep and cognitive

variables as the dependent variables and time (linear, quadratic, and cubic functions) as the

independent variables. The subsequent unstandardized residual values resulting from the

regressions were then saved and used as time independent values. Using the residual values

computed above, an index of between-person variability (Sample Standard Deviation, SD) and

within-person variability (Individual Standard Deviation, ISD) will be computed. These values

will then be compared by dividing the ISD by the SD to get the proportion of between-person

variability that is found within-persons.










Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive Measures
in Older Adults with Insomnia

Importance

If within-person variability operates systemicallyy", then the tendency to be variable in one

domain should be associated with variability in other domains. In other words, are people

"generally stable" or "generally inconsistent"? If increased within-person variability is related to

increased or decreased performance then variability can be labeled as either flexibility or

vulnerability.

Hypothesis

Several studies (e.g., Allaire & Marsiske, 2002) have shown at least modest correlations

among the within-person variability estimates across persons and have suggested that variability

is related to increased performance. How sleep variability relates to cognitive variability in aged

insomniacs and whether this variability is 'good' is unknown. However, it is believed that

variability will be related across constructs and that variability will be related to increased

performance .

Analysis

To determine how within-person variability is related across constructs, bivariate

correlations among intraindividual variability estimates (residualized ISDs to control for time

effects) across measures will be run. To determine how mean-level performance is related across

constructs, bivariate correlations among mean-level values will be run. To determine how

within-person variability relates to average performance, bivariate correlations between

intraindividual variability estimates (residualized ISDs to control for time effects) and mean

values will be run. An ISD analysis produces a "trait-like" inconsistency score for each









individual. Such analysis addresses questions regarding who is more or less consistent and what

the antecedents and consequences of any between-person differences in inconsistency might be.

Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-to-Day-
Level) Association between Sleep and Cognition in Older Adults with Insomnia

Importance

All prior research has examined the between-person association between sleep and

cognition. There are no published data regarding the day-to-day, within-person, associations

between sleep and cognition, especially in older adults with insomnia. However, if day-to-day

associations between sleep and cognition are found, the importance of a good night' s sleep for

optimal cognitive functioning would be supported.

Hypothesis

In general it is expected that on average better sleep will relate to better cognitive

performance (between-persons, Level 2 Fixed Effect) and days with better sleep will relate to

days of better cognitive performance (within-person, Level 1 Fixed Effect). Specifically,

spending long amounts of time falling asleep and awake in the middle of the night (on average

and daily) will relate to poorer executive functioning and poorer processing speeds. Hypothesis

about the relationship between terminal wakefulness and cognitive performance can not be made

due to the lack of prior research on this sleep variable.

Analysis

The current aim is to examine the predictive power of within-person and between-person

sleep variables on cognition performance. To accomplish this, daily data from the obj ective and

subj ective sleep measures (SOL, WASO, TWAK) will be used to predict cognitive functioning

(processing speed and executive functioning) applying a multilevel model (MLM) approach.

MLM, also referred to as mixed effects modeling or hierarchical linear modeling (HLM; Bryk &









Raudenbush, 1992), is an extension of the general linear model, and does not require

observations to be independent. Thus, MLMs are very flexible and especially suited for daily

data because of their autoregressive nature and hierarchical structure with daily observations

nested within each participant (Singer, Davidson, Graham, & Davidson, 1998; Singer, Fuller,

Keiley, & Wolf, 1998; Singer & Willett, 2003).

Because of the hierarchical nature of our data (14 consecutive days nested within 48

participants) and in order to increase the precision of predicting fluctuations in processing speed

and executive functioning with changes in sleep patterns, we will model the data with a MLM

approach. This provides the opportunity to examine how well sleep predicts cognition both

within- (level 1) and between- (level 2) persons. Level 1 analysis addresses questions such as:

"On days in which a person reports above-average sleep onset latencies, does s/he also

experience lower levels of processing speed? This level of analysis is concerned with questions

of atypical days within an individual and what predicts, within-persons, the consequences of

these atypical days. Level 2 analyses examines questions like: "Do people who are generally

better sleepers report higher levels of processing speed?"

Both obj ective and subj ective sleep measures will be used to predict cognitive

functioning (processing speed and executive functioning) using a five-step MLM approach. Step

1 (Table 3.1, Row 1), the null (baseline) model, will estimate only a fixed and random intercept

for cognitive functioning (Bryk & Raudenbush, 1992). This model will specify that cognition for

person j on day i is a function of the overall group-average cognition (yoo), a between-person

random error term (uo,), and a within-person random residual component (et). This step provides

a comparison for later models.









In step 2, time functions (linear) will be added as a covariate to the null model (Table 3.1,

Row 2), producing a latent growth curve model. As such, the model will specify that cognition

for person j on day i is a function of: average cognitive level (Too), linear time (Pe), a between-

person random error term (uo,), and a within-person random residual component (et). This step

controls for any within-person inflations that may be caused by a systematic growth in the data.

Next, measures of subj ective and obj ective sleep will be added to the model. In step 3

(Table 3.1, Row 3) the estimates of the fixed and random intercepts and fixed linear slopes for

each sleep variable will be added. Thus, the daily cognition scores (Cognition,,) for each person

will be predicted by: average level of cognition (Too), linear time (Pe), the between-person effects

of mean-level subj ective and obj ective sleep scores, the within-person effects of daily-centered

subj ective and obj ective sleep scores, a between-person random error term (uo,), and a within-

person random residual component (eti).

In step 4 (Table 3.1, Row 4), level 1 level 2 interaction terms of like variables (e.g.,

level 1 SOLs level 2 SOLs) will be entered to estimate the effect of within-person fluctuations

in sleep for individuals who on average sleep more-or-less than others.

In step 5 (Table 3.1, Row 5), the random linear slopes (u,) of the significant daily-centered

subj ective and obj ective sleep variables and interactions will added in order to estimate any

between-person differences in the prediction of cognition.

All models will be estimated using the Maximum Likelihood (ML) method. The ability of

a model to predict cognition better than the baseline model (i.e., Deviance) will be used as an

index of Goodness of Fit. Improvements in predictability will be determined by the amount of

reduction of within- and between-person residual variances compared to the baseline model

(Bryk & Raudenbush, 1992). Decreases in residual and intercept variances represent a









proportional reduction of the prediction error, which is analogous to R2, and will be used as an

estimates of within and between-person effect sizes. The amount of agreement between model

predicted values and actual values will be calculated as an estimate of an overall effect size.

Because this study assessed sleep repeatedly, and used two different methods (sleep

diaries and actigraphy), issues of multicollinearity will be a concern. Formal multicollinearity

diagnostic procedures are not available for multilevel modeling. To assess for possible

collinearity, a multivariate mixed-effects null model will be estimated. This procedure will

produce both "G" (between-persons) and "R" (within-persons) covariance matrices, which will

be subsequently rescaled into correlations using the following equation (Equation 3-1):

(3-1) Correlationx,y COVariance termx,y i lVaricrCct x "VariaHCc .











Table 3 -1. Level 1 and Level 2 Equations at each Step, of the MLM Building. Process
Step> Level 1 equation Level 2 equation
1Cognitionij = po, + e,, oi = Yoo+ Mo
2 Cognitionij = po, + PTime e, Po = Yoo+ uo,

3 Cognitionij = Po, + p y,Time + 2/(SOL~o!, P, =, 7oo Y SO~Lo p y02 SOL~s, +yosr WASOol
SOLo,) P3;(SOLs, SOLs,) P4/(WASOo, +Y04 WASOs, yY os TWAKo, +y o TW4Ks,+ uo,
WASOo,) PS,(WASOs, WASOs,) .
P6,(TWAKo, TW4Ko,)+ P,(TWAKs,
TW4Ks ,) e,
4 Cognitionij = Bo, + P yTime $2/(SOLo,- Ipy = 7o+ rio; SOI'oI Y02 SOL~s, + Yo ASOo,
SOLo,) P3;(SOLs, SOLs,) P4/(WASOo, +Y O4 WASOs, y. o TW4Ko,. +y TW4Ks,, uo,
WASOo,) PS,(WASOs, WASOs,) .
P6,(TWAKo, TW4Ko,) P7,(TWAKs,
TW4Ks,), P,[ (SOLo,- SOLo,) SOLo,].
P9,[(SOLs,- SOLs,) SOLs,].
p,,[(WASOo, WASOo,) WASOo,]
P;,[ (WASOs, WASOs,) WASOs ,]+
1,~[ (TWAKo, TW4Ko,) TW4Ko,]
P13/[ W~AKsll MWAKs,)" AKs,]. e,
5 Cognitionij = po, + p yTime + 2/(SOLo,- De = Yoo+ Yo;i SOLo,+ Y0 SOLs, + Yos WAS~Oo
SOLo,) P3;(SOLs, SOLs,) P4/(WASOo, +Y04 WASOs, yY os TW4Ko, +y o TW4Ks,+ uo,
WASOo,). PS,(WASOs, WASOs,) + 02=Ylo+ u;
P,(TWAKo, TW4Ko,). P,(TWAKs, I 27=Y20+ u2;
TW4Ks,) Ps,[ (S OLop SOLo,) SOLo,] 3= 0+ 3
Pg[(SOLs, SOLs,) SOLs,]. Ps,=Ysous,
Plos[(WAS~oo pWASOo,) WASOo,]. P6-Y6o+ ug
P;,~[(WASOs, WASOs,) WASOs ,]. Ps= Y ous
Pl,~[(TWAKo, TW4Ko,) TW4Ko,]. s=s~s
ss= Yso+ us
Pl3/[(TWAKs!, TW4Ks,) TW4Ks,]. e,, Pio;=y~iw+Mzo
11;=r110+ 11;
12/~=Y120+u121
1= 130+ U13
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a sub script 'o' indicate they were obj ectively measured by Actigraphy. SOL =
sleep onset latency, WASO = wake after sleep onset, TW4K = terminal wakefulness.











CHAPTER 4
METHOD S

Main Study

Data presented in this document were obtained during a two-week baseline assessment of

sleep and cognitive performance in the initial phase of a randomized, controlled trial of the

cognitive effects of a cognitive-behavioral intervention for older adults with insomnia (Research

and Education on Sleep and Thinking study, REST; McCrae, PI; AGO24459-01). Older adults

with insomnia were recruited from North central Florida by means of newspaper, radio, and

television advertisements. At the time of analysis forty-eight individuals completed baseline

assessment and were included in the present study.

As stated previously, REST is a randomized, controlled trial. The study included two

weeks of baseline (followed by randomization), four weeks of either waitlist control or insomnia

treatment, two weeks of post-testing, and a two week follow-up at three months. Through each

phase of the study participants completed daily sleep and cognitive measures. The overall design

of the REST Study is outlined in figure 4. 1.

Participants

Participants were recruited for participation in a clinical trial of the possible cognitive

benefits of a cognitive-behavioral treatment for late-life insomnia. Potential participants

responded to advertisements offering a free non-drug treatment for insomnia for individuals sixty

years of age and older. To ensure their suitability for the study, participants went through a

thorough screening process that included many inclusion and exclusion criteria.

Inclusion Criteria

To be included in the study potential participants had to be sixty years old or older and

meet the diagnostic requirements of primary insomnia. These diagnostic requirements include:









(1) self-report sleep onset latency or wake after sleep onset greater than thirty minutes a night,

(2) sleep disturbance present at least three nights a week for six months or greater, and (3) a

daytime dysfunction due to the insomnia (social, occupational, mood, or cognitive) (American

Academy of Sleep Medicine, 2005; American Psychiatric Association, 1994). Furthermore,

participants had to be either stabilized on any sleep promoting substance for at least six months

or not have taken any such substance for the past month.

Exclusion Criteria

Participants were excluded from the study if they reported any significant medical (e.g.,

stroke), neurological (e.g., Parkinson's disease), psychological (e.g., schizophrenia, bipolar

disorder) or sleep disorder other than insomnia (e.g., sleep apnea, periodic limb movements).

Participants were also excluded if they were found to have severe depression as indicated by a

score of twenty-four or higher on the Beck Depression Inventory 2nd Edition (BDI-II; Beck,

Steer, & Garbin, 1996) or a score of thirteen or higher on the Geriatric Depression Scale (GDS;

Yesavage, 1983). To screen for potential Alzheimer' s disease/dementia a cutoff score of lower

than twenty-three, for individuals with a 9th grade education and higher, and a score of eighteen

and lower, for individuals with less than a 9th grade education, was implemented (MMSE;

Folstein, Folstein, & McHugh, 1975). The use of psychotropic or other medication known to

affect sleep, such as beta-blockers, also excluded individuals from participating in the study.

Sample Characteristics

The sub-sample available at the time of the analysis included forty-eight older insomniacs.

The sample can be categorized as "young-old" as the mean age at the time of the study was

69.91 years (SD = 7.24). The sample was highly educated (M = 16. 17 years, SD = 2.98 years),

mostly married (54.3% married) and female (62.5 % female), and Caucasian (95.7% first

language English). The sample appeared to be relatively healthy, reporting to take an average of









3.41 prescription medications (SD = 2.50) and 2.17 over-the-counter medications (SD = 1.92).

The sample reported to be chronic insomniacs, with an average of 13.96 years of insomnia

symptoms (SD = 14.54). Furthermore, participants reported spending an average of 133.67

minutes awake per night (SD = 100.89) and only sleeping 360.57 minutes a night (SD = 126.17).

See Table 4-1 for a complete list of the sample' s sleep characteristics.

Measures

Sleep Measures

All sleep variables were measured concurrently with a subj ective measure, daily sleep

diaries (Lichstein, Riedel, & Means, 1999), and an objective measure, wrist-worn actigraphy

(Mini Mitter Co., 2001), for fourteen consecutive days during the baseline of the REST Study.

Because all of the sleep variables were measured in two distinct ways (sleep diaries and

actigraphy), an 's' subscript indicates the variable was measured subjectively, and an 'o'

subscript indicates it was measured obj ectively.

Subjective sleep measures

Participants completed sleep diaries (Lichstein et al., 1999) each morning for 14 days,

providing subjective estimates of the following sleep- wake parameters: (1) sleep onset latency

(SOLs)-estimated by participants as the time it took them to fall asleep after laying down; (2)

wake time after sleep onset (WASOs)- estimated by participants as the total amount of time

spent awake during the night; and (3) terminal wakefulness (TWAKs), computed by subtracting

final wake-up time (actual time the participant awoke in the morning) from out of bed time (time

participant actually got out of bed). For reference purposes, a sleep diary is reproduced in

Appendix A.










Objective sleep measures

Participants wore an actigraph, the Actiwatch-L, which has an integral ambient light sensor

(Mini Mitter Co., 2001), on their nondominant wrist for 14 consecutive days, concurrent with the

sleep diary period. The Actiwatch-L monitors ambient light exposure and gross motor activity

and contains an omni-directional, piezoelectric accelerometer with a sensitivity of > 0.01 g-force

and a light sensor with a recording range of 0. 1 to 150,000 Lux.

The sensors of the Actiwatch-L are sampled 32 times /second and record the peak value for

each second. These peak values are then summed into 30-second "activity" counts. These

activity counts are then downloaded to a PC and analyzed using Actiware-Sleep v. 3.3 (Mini

Mitter Co., 2001), which uses a validated algorithm to identify each epoch as either sleep or

wake. The software provides three default sensitivity settings (high, medium, low). This study

utilized medium sensitivity. On medium sensitivity, the threshold is set at 40 activity counts. If

the total activity for an epoch was > 40, it was scored as wake. If the total activity was < 40, the

final activity count for the epoch was based on the level of activity in the surrounding 2 minutes

(see Equation 4-1).

(4-1) Total Activity

Epoch A = EA-4 (.04) + EA-3 (.04) + EA-2 (.20) + EA-1 (.20) + E (2) + EA 1 (.20) + EA 2 (.20) + EA+3 (.04) + EA 4 (.04),

Where A = activity count for the epoch being scored and EA +/- 1-4 = activity count in adj acent
epochs. If Epoch A Total Activity (i.e., weighted sum of activity counts) exceeded the threshold
of 40, then Epoch A was scored as wake; otherwise, it was scored as sleep.

Bedtime and time out of bed in the morning were based on sleep diary entries as

recommended in the software manual (Mini Mitter Co., 2001). Actiware-Sleep determines sleep

start automatically by searching for the first 10 minutes during which no more than one epoch

scored as wake. Likewise, sleep end was the last 10 minutes during which no more than one

epoch scored as wake. As previously mentioned, Actiware-Sleep provides objective estimates









for all of the variables also provided by sleep diaries. Those variables and their definitions when

measured obj ectively by actigraphy are: (1) SOLo-amount of time from lay down to first sleep

epochs; (2) WASOo-sum of all wake epochs after first sleep epoch; and (3) TWAKo-amount of

time from last sleep epoch to get out of bed time.

Cognitive Measures

Cognition was measured in two broad domains: processing speed and reasoning. Both

cognitive domains were assessed daily via paper and pencil, self-administered tasks for the

fourteen consecutive days of baseline during the REST Study.

Reasoning

Inductive reasoning reflects the ability to infer general principles from specific instances

and apply these general principles to new instances of the problem. Inductive reasoning is highly

related to working memory (Salthouse, 1991) and to components of executive functioning

(Lezak, 1995).

The letter series task (Thurstone, 1962) was used to measure inductive reasoning. In this

task, participants have to identify the pattern for a series of letters. Participants are asked to

choose the letter that would continue the established pattern (A BD AB D AB _?) in a series

of letters from five answer choices. Participants are given four minutes to complete as many

items as possible. The maximum score is 30. The performance score is the number of correct

responses. Over the course of the study fourteen alternate forms were given, one per day. These

alternate versions of the letter series test have been shown to have high test-retest reliabilities

across four blocks, each containing fifteen days of assessment (Allaire & Marsiske, 2005). See

Appendix B for an example of the letter series test.









Processing speed

The Symbol Digit Modalities Test (Smith, 1982), which measures perceptual speed and

processing speed was used. In this test, the participant is presented with a series of nine symbols

that are paired with a single, unique digit in a key at the top of an 8.5 x 11 inch sheet. The rest of

the page displays a randomized sequence of symbols with blank spots below each. The

participant is instructed to write the digit that corresponds to each symbol in the key below that

symbol as quickly as possible. The participant is given 90 seconds to place as many correct digits

below their corresponding symbol. The outcome of interest is the total number of correct

responses recorded in the allotted 90 seconds.

The SDMT has many applications and has been administered to individuals of many

different cultures, age and education levels (Lezak, 1995). It has been administered both

individually and in groups. The SDMT has been used in the detection of dyslexia, aphasia and/or

cerebral dysfunction and for early screening of normal elders for manual motor defects, visual

acuity, oculomotor coordination, and visuo-spatial orientation difficulties (Smith, 1991). Test-

retest reliability, alternate form reliability, and convergent validity for the SDMT are all at or

above .80 (Lezak, 2004). Over the course of the study fourteen alternate forms of the SDMT

were used, one per day. These fourteen alternate versions of the symbol digit test have been

shown to have high test-retest reliabilities across thirty days of assessment (McCoy, 2004). See

Appendix C for a reproduction of the SDMT.

Procedures

Baseline Study Procedure

Study procedures from the REST Study baseline period that are relevant to the present

study include: an initial telephone interview, an overnight portable polysomnography (PSG;

Medcare Diagnostics) assessment, and two in-laboratory visits where actigraphic sleep









measurement, subjective sleep measurement with sleep diary, and daily cognitive workbooks

were collected.

An initial telephone interview was conducted with all interested potential participants.

During this half-hour phone interview, demographic information was collected, insomnia

symptoms were assessed, medical history was taken, and an at-home visit was scheduled.

At the in-home visit participants were instructed in the operation of the ambulatory PSG.

The PSG measures blood-oxygen saturation and respiration during sleep and was used to rule out

sleep apnea as the cause of the participants sleep complaints. All participants were instructed to

wear the PSG throughout the night and bring the unit in to the laboratory the following day for

reading. An apnea-hypoxia index (AHI) of 13.1 and higher was used as the cutoff for

participation in the study.

During the first laboratory visit participants were instructed in the use of the actiwatch,

sleep diaries, and daily cognitive workbooks (which contained the symbol digit test and letter

series test). Participants were instructed to wear the actiwatch twenty-four hours a day and to fill

out the sleep dairy daily upon awakening. A sample cognitive workbook was explained and

participants were told to fill out one daily in the morning. A week from this initial meeting

participants returned at which time their actiwatch data was downloaded and their sleep diary

and seven cognitive workbooks were collected. The participants were then given new material

and scheduled for an appointment to come back in a week and turn in the material again.

Alternate Forms of the Cognitive Measures

During the fourteen days of baseline assessment participants completed both the symbol

digit tests and letter series test daily. In an attempt to minimize practice effects commonly found

in repeated cognitive assessments, fourteen alternate forms of each test were used. The alternate









forms were constructed to be comparable in difficulty and cognitive resources needed to

complete them.

The alternate forms of the symbol digit tests were constructed by changing the pairings of

symbols to digits in each form such that no two forms of the test consisted of a digit-symbol

pairing that appeared in any other form. This manipulation was done to prevent participants from

memorizing the digit-symbol pairing.

The alternate forms of the letter series test were constructed by changing the start letter of

all letter patterns but maintaining the underlying pattern concept (e.g., AA BA AC AA DA A

would become DD E DD F D DG D D). Pattern length and number of distracters in the answer

choices was maintained across all forms. No two forms of the test contained identical questions.

Missing Data

Throughout the two week baseline measurement period some participants forgot to wear

their actiwatch for a day or fill out their sleep diary on a given day or do one or both of the

cognitive tasks. Because analyses was done using multilevel modeling (MLM), also know as

hierarchical linear modeling and mixed effects modeling, missing data is not an issue. MLM

allows all available data to be included in analyses, as it assumes random missing data (Bryk &

Raudenbush, 1992) and therefore does not exclude whole cases (known as case-wise exclusion)

due to one missing data point.















Randomization ~ olw



Two eeksTwo Weeks Two Wee


Four Weeks

Figure 4-1. Overall design of REST Study. Boxed in portion, baseline, is where all study
material was obtained.


:ks













Table 4-1. Participant Characteristics
N Minimum Maximum Mean Std. Deviation
Actigraphy
Sleep Onset Latency 48 0.00 351.00 20.60 32.74
Wake After Sleep Onset 48 0.00 154.00 31.92 20.99
Terminal Wakefulness 48 0.50 211.00 18.07 25.79
Total Wake Time 48 0.00 154.00 31.94 21.01
Total Sleep Time 48 118.50 659.50 424.54 82.52
Sleep Efficiency 48 35.90 99.92 85.80 9.15
Sleep Diary
Sleep Onset Latency 48 0.00 345.00 39.04 46.29
Wake After Sleep Onset 48 0.00 550.00 58.08 60.70
Terminal Wakefulness 48 0.00 270.00 31.43 42.43
Total Wake Time 48 7.00 660.00 127.80 100.23
Total Sleep Time 48 0.00 700.00 360.95 117.86
Sleep Efficiency 48 0.00 98.50 73.59 20.55
Note: All values measured in minutes.









CHAPTER 5
RESULTS

The overarching aim of the study was to examine the intraindividual variability in

obj ective and subj ective sleep and cognitive performance in older adults with insomnia. The

results are grouped and presented by the three main obj ectives of the study:

1. How much variability in sleep and cognition in older adults with insomnia is found
within-persons and how much is found between-persons?

2. How is variability related across the various sleep and cognition domains?

3. What are the between-person and within-person associations between sleep and
cognition in older adults with insomnia?

Aim 1: To Determine the Amount of Variability in Sleep and Cognition Found Within
Older Adults with Insomnia Compared to the Amount of Variability between Older Adults
with Insomnia

The first goal of the study was to demonstrate that there is a considerable amount of

within-person variability in both sleep and cognition in older adults with insomnia. Dissection of

the sleep and cognition variables did reveal a considerable amount of variability found within-

persons. In fact, all but one variable, letter series, was found to display at least 50% of the

amount of between-person variability within-persons (Although the letter series test did not

exhibit 50% of the amount of between-person variability, it did display 48% of the amount of

between-person variability within-persons). All other variables did display at least the

hypothesized 50% of between-person variability within-persons.

The number correct on the symbol digit test was found to vary within-persons 92% as

much as it varies between-persons. Subj ectively measured sleep was also found to vary

considerably within-persons. SOLs displayed 85% of the amount of between-person variability

within-participants; WASOs displayed more within-person variation (113%) than between-

person; and, TWAKs exhibited a quarter more fluctuation within-person (125%) than between-










person. All obj ectively measured sleep variables displayed more variability from day-to-day,

within-persons, than between-persons. Variability within-persons in SOLo was found to be

nearly a third larger (132%) than between-persons; WASOo displayed 1 12% of the amount of

between-person variability within-persons; and, TWAKo exhibited nearly double the between-

person variability within-persons (189%). For a completed listing of the amount of within-person

variability compared to between-person variability, see Table 5-1. For a graphical depiction of

the relative amount of within-person variability, see Figure 5-1.

Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive Measures
in Older Adults with Insomnia

The second obj ective of the study was to determine how variability is associated across the

various sleep and cognition domains. Results for this aim will be presented in three parts. The

first part will consider associations between within-person variability indices across sleep and

cognition. The second part will examine the mean-level associations across sleep and cognition.

And, the third part will examine the associations between within-person and mean-level indices

of sleep and cognition.

Associations of Within-Person Variability

Correlations of the within-person standard deviations (residualized ISDs to control for time

effects) revealed relationships between within-person variability in sleep and cognition that

ranged from 0.04 to 0.40. Increased within-person variability in SOLs was significantly and

positively associated with increased variability in the symbol digit test (r = 0.40, p < .01) and

increased within-person variability in TWAKs was significantly and positively associated with

increased within-person variability in the letter series test (r = 0. 14, p < .01) and symbol digit test

(r = 0.35, p < .01). Increased within-person variability in SOLo was significantly and positively

related to increased within-person variability in the symbol digit test (r = 0.36, p < .01). Within-










person variability in WASOo and TWAKo were both significantly and positively associated with

within-person variability in the letter series test (r = 0.26, p < .01; r = 0.26, p < .01) and the

symbol digit test (r = 0.31, p < .01; r = 0.37, p < .01). Within-person variability in all sleep

variables were significantly and positively associated with within-person variability in all other

sleep variables (r's = 0.08 0.62, all p' s < .05). These results suggest that inconsistency operates

in a trait-like fashion within individuals and across sleep and cognition. See Table 5-2 for a

complete listing of within-person variability correlations.

Associations of Mean-Level Performance

Correlations of mean-level sleep and cognition revealed associations that ranged from

-0.06 to 0.27. Significant negative associations between mean-level sleep and number correct on

the letter series test were found for SOLs (r = -0. 14, p < .01), TWAKs (r = -0. 12, p < .01), SOLo

(r = -0.09, p < .05), and WASOo (r = -0. 16, p < .01). A significant positive association was found

between mean-level TWAKo and letter series (r = 0. 12, p < .01). Significant positive

associations between mean-level sleep and number correct on the symbol digit test were found

for SOLs (r = 0.27, p < .01), TWAKs (r = 0.22, p < .01), SOLo (r = 0. 15, p < .01), WASOo (r =

0. 12, p < .01), and TWAKo (r = 0. 18, p < .01). Associations among the mean-level indices in

sleep ranged from 0.04 to 0.61. Results indicate that increased unwanted wake time is associated

with increased processing speed and decreased reasoning. See Table 5-3 for a complete listing of

mean-level correlation coefficients.

Associations between Within-Person Variability and Mean-Level Performance

Correlations of mean-level sleep and within-person variability in cognition ranged from

0.03 to 0.45. Significant positive associations between mean-level sleep and within-person

variability on the number correct on the letter series test were found for SOLs (r = 0.24, p < .01),

WASOs (r = 0.09, p < .05), TWAKs (r = 0.30, p < .01), WASOo (r = 0. 18, p < .01), and TWAKo










(r = 0.21, p < .01). Significant positive associations between mean-level sleep and within-person

variability on the number correct on the symbol digit test were found for SOLs (r = 0.39, p <

.01), TWAKs (r = 0.3 5, p < .01), SOLo (r = 0.3 8, p < .01), WASOo (r = 0.24, p < .01), and

TWAKo (r = 0.45, p < .01). These results indicate that higher levels of wake time are associated

with more variable cognitive functioning.

Significant negative associations between within-person variability in sleep and mean-level

performance on the letter series test were found for SOLs (r = -0.13, p < .01), TWAKs (r = -0.25,

p < .01), and SOLo (r = -0.10, p < .05) indicating that inconsistency in sleep is negatively related

to level of reasoning. Significant positive associations between within-person variability in sleep

and mean-level performance on the symbol digit test were found for SOLs (r = 0.28, p < .01),

TWAKs (r = 0.23, p < .01), and TWAKo (r = 0.09, p < .05) indicating the inconsistency in sleep

is positively associated with level of processing speed. Associations between within-person

variability in sleep and mean-level sleep ranged from 0.04 to 0.88, indicating that increased

variability in sleep is associated with worse overall sleep. Associations between inconsistency in

cognitive performance and level of performance indicated that both variability in reasoning and

processing speed are associated with better reasoning (r = 0.34, p < .01) and better processing

speed (r = 0.53, p < .01). Please see Table 5-4 for a complete listing of associations between

within-person variability indices and mean-level indices across sleep and cognition.

Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-to-Day-
Level) Association between Sleep and Cognition in Older Adults with Insomnia

Multicollinearity

Prior to running the multilevel models to predict letter series and symbol digit performance

a series of within-person and between-person correlations was run among predictors variables to









determine the extent to which the independent variables shared variances. Formal

multicollinearity diagnostic procedures are not available for multilevel modeling.

Within-person multicollinearity

The within-person correlational analysis revealed significant collinearity between SOLs

and TWAKs (r = 0. 10, p < .05), between WASOo and WASOs (r = 0. 14, p < .01) and TWAKs (r

= 0. 11, p < .05) and SOLo (r = -0. 10, p < .05), and between TWAKo and TWAKs (r = 0. 16, p <

.01) and SOLo (r = 0. 15, p < .01) and WASOo (r = -0. 12, p < .01). See Table 5-5 for a complete

listing of within-person correlation estimates. For a listing of within-person correlations among

predictor variables when not controlling for time see Appendix D.

Between-person multicollinearity

The between-person correlational analysis revealed significant collinearity between

WASOs and SOLs (r = 0.48, p < .01), and between TWAKs and SOLs (r = 0.45, p < .05) and

WASOs (r = 0.48, p < .01), and between SOLo and SOLs (r = 0.74, p < .01), and between

TWAKo and SOLo (r = 0.44, p < .05) and WASOo (r = 0.50, p < .01). See Table 5-6 for a

complete listing of between-person correlation estimates. For a listing of between-person

correlations among predictor variables when not controlling for time see Appendix D.

Multilevel Model for Letter Series

The intraclass correlation coefficient (ICC), which serves as an index of the amount of

within and between-person variability to be explained (Bryk & Raudenbush, 1992), was 0.3 6.

This indicates that, 64% of the overall variability in letter series was within-person and 36% was

between-person. For a complete listing of model parameters and estimates obtained at each step

of the model building process see Table 5-7.

In the final MLM for letter series performance TWAKo was the only significant between-

person, level 2, predictor, P = 0.15, t(45.43) = 2.03, p = .05. At the within-person level, level 1,










both the predictors of Day, P = 0.37, t(46.87) = 8.53, p < .001, and SOLs, P = 0.01, t(494.50) =

1.94, p = .05, were significant. A significant interaction was found between level 2 and level 1

SOLs, P = 0.00018, t(482.3 8) = -2.42, p < .05. The model also contained a significant random

effect of Day, P = 0.04, Wald's Z = 2.73, p = .01.

This model explained approximately 3 5% of the within-person variance and 25% of the

between-person variance. The model accounted for roughly 86% of the total variance in letter

series performance. See Table 5-8 for a total listing of predictor estimates and significance

levels. See Appendix E for total listings of predictor estimates and significance levels for MLMs

estimates in model building steps 2-4.

Multilevel Model for Symbol Digit

The intraclass correlation coefficient (ICC) for symbol digit was 0.73. This indicates that,

27% of the overall variability in symbol digit was within-person and 73% was between-person.

For a complete listing of model parameters and estimates obtained at each step of the model

building process see Table 5-9.

In the final MLM for symbol digit performance, WASOs was the only significant

between-person, level 2, predictor, P = -0.08, t(1.47) = -2.08, p = .04. At the within-person level,

level 1, only the predictor of Day, P = 0.48, t(144.30) = 3.39, p = .001, was significant. A

significant interaction was found between level 2 and level 1 TWAKs, P = 0.001, t(465.84) =

2. 13, p < .05, and level 2 and level 1 TWAK 0, p = 0.004, t(22.62) = 2.26, p < .05. The model

contained no significant random effects.

This model explained approximately 1% of the within-person variance and 23% of the

between-person variance. The model accounted for roughly 44% of the total variance in symbol

digit performance performance. See Table 5-10 for a total listing of predictor estimates and










significance levels. See Appendix F for total listings of predictor estimates and significance

levels for MLMs estimates in model building steps 2-4.












Table 5-1. Amount of within and between-person variability
Sample Standard Deviation (SD) Individual Standard Deviation (ISD) SD / ISD
Letter Series 4.96 2.38 0.48
Symbol Digit 9.05 8.36 0.92
Sleep Onset Latency, 31.58 26.86 0.85
Wake After Sleep Onset, 37.77 42.71 1.13
Terminal Wakefulness, 25.03 31.32 1.25
Sleep Onset Latencyo 16.39 21.71 1.32
Wake After Sleep Onseto 13.09 14.72 1.12
Terminal Wakefulnesso 10.86 20.58 1.89
Notes: All values shown have been de-trended for any linear or quadratic effects of time. Variables with a
subscript 's' indicate they were subjectively measured by Sleep Diary. Variables with a subscript 'o'
indicate they were objectively measured by Actigraphy.

















0




See DiryAtirah

Fiue51. eaieaon fwti-esnvraiiycmae obtenpro aiblt

afe cotoligf~or nlna orO qudrtc fecsoftme













Table 5-2. Correlations of within-person standard deviations (N = 48)
1 2 3 4 5 6 7 8
1. Letter Series 1.00
2. Symbol Digit 0.20** 1.00
3. Sleep Onset Latency, 0.06 0.40** 1.00
4. Wake After Sleep Onsets -0.04 -0.06 0.45** 1.00
5. Terminal Wakefulness, 0.14** 0.35** 0.53** 0.28** 1.00
6. Sleep Onset Latencyo 0.04 0.36** 0.35** 0.08* 0.14** 1.00
7. Wake After Sleep Onseta 0.26** 0.31** 0.12** 0.24** 0.17** 0.42** 1.00
8. Terminal Wakefulnesso 0.26** 0.37** 0.31** 0.30** 0.26** 0.33** 0.62** 1.00
Note: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level
(2-tailed). Variables with a subscript 's' indicate they were measured subjectively by Sleep Diary.
Variables with a subscript 'o' indicate they were objectively measured by Actigraphy. All values have
been de-trended to control for any linear or quadratic effects of time.













Table 5-3. Correlations of sample means (N = 48)
1 2 3 4 5 6 7 8
1. Letter Series 1.00
2. Symbol Digit 0.23** 1.00
3. Sleep Onset Latency, -0.14** 0.27** 1.00
4. Wake After Sleep Onsets -0.06 -0.06 0.43** 1.00
5. Terminal Wakefulness, -0.12** 0.22** 0.41** 0.41** 1.00
6. Sleep Onset Latencyo -0.09* 0.15** 0.61** 0.09* 0.21** 1.00
7. Wake After Sleep Onseta -0.16** 0.12** 0.14** 0.22** 0.18** 0.04 1.00
8. Terminal Wakefulnesso 0.12** 0.18** 0.27** 0.26** 0.29** 0.32** 0.32** 1.00
Notes: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level
(2-tailed). Variables with a subscript 's' indicate they were subjectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were objectively measured by Actigraphy. All values have
been de-trended to control for any linear or quadratic effects of time.












Table 5-4. Correlations of within-person standard deviations and sample means (N = 48)
1. (SD) 2. (SD) 3. (SD) 4. (SD) 5. (SD) 6. (SD) 7. (SD) 8. (SD)
1. Letter Series (M) 0.34** -0.15** -0.13** 0.00 -0.24** -0.10* -0.07 -0.05
2. Symbol Digit (M) 0.34** 0.53** 0.28** -0.08 0.23** 0.02 0.00 0.09*
3. Sleep Onset Latency (M), 0.24** 0.39** 0.77** 0.21** 0.45** 0.59** 0.29** 0.35**
4. Wake After Sleep Onset (M), 0.09* 0.03 0.55** 0.60** 0.43** 0.13** 0.17** 0.22**
5. Terminal Wakefulness (M), 0.30** 0.35** 0.22** 0.12** 0.69** 0.15** 0.15** 0.20**
6. Sleep Onset Latency (M)o 0.06 0.38** 0.30** 0.05 0.22** 0.88** 0.35** 0.28**
7.Wake After Sleep Onset (M)o 0.18** 0.24** 0.09* 0.04 0.24** 0.07 0.59** 0.32**
8. Terminal Wakefulness (M)o 0.21** 0.45** 0.23** 0.29** 0.25** 0.32** 0.58** 0.89**
Notes: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed). Variables with a subscript 's'
indicate they were subjectively measured by Sleep Diary. Variables with a subscript 'o' indicate they were objectively measured by Actigraphy.
All values have been de-trended to control for any linear or quadratic effects of time. (M) indicates the value is the sample mean. (SD) indicates
the value is the within-person standard deviation. All values have been de-trended to control for any linear or quadratic effects of time.












Table 5-5. Correlations of within-person predictors (N = 48, 14 Occasions)
1 23 4 5 6 7 8
1. Letter Series 1.00
2. Symbol Digit 0.08 1.00
3. Sleep Onset Latency, 0.00 0.00 1.00
4. Wake After Sleep Onset, 0.05 0.03 -0.04 1.00
5. Terminal Wakefulness, 0.07 0.03 0.10* 0.05 1.00
6. Sleep Onset Latencyo -0.01 0.01 0.06 0.01 0.01 1.00
7. Wake After Sleep Onseta -0.01 0.04 0.08 0.14** 0.11** -0.10* 1.00
8. Terminal Wakefulnesso 0.07 0.04 -0.05 0.04 0.16** 0.15** -0.12** 1.00
Notes: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level
(2-tailed). Variables with a subscript 's' indicate they were subjectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were objectively measured by Actigraphy. All values have
been de-trended to control for any linear or quadratic effects of time.












Table 5-6. Correlations of between-person predictors (N = 48).
1 2 3 4 5 6 7 8
1. Letter Series 1.00
2. Symbol Digit 0.23 1.00
3. Sleep Onset Latency, -0.14 0.30 1.00
4. Wake After Sleep Onsets -0.06 -0.07 0.48** 1.00
5. Terminal Wakefulness, -0.15 0.25 0.45* 0.48** 1.00
6. Sleep Onset Latencyo -0.11 0.21 0.74** 0.11 0.29 1.00
7. Wake After Sleep Onseta -0.19 0.14 0.16 0.26 0.18 0.08 1.00
8. Terminal Wakefulnesso -0.14 0.25 0.41 0.34 0.39 0.44* 0.50** 1.00
Notes: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level
(2-tailed). Variables with a subscript 's' indicate they were subjectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were objectively measured by Actigraphy. All values have
been de-trended to control for any linear or quadratic effects of time.











Table 5-7. Steps taken in building the Letter Series Multilevel Model
Letter Series
Models AIC BIC -2LL A-2LL dfAf s2b S2w ~ b w t
(1) Null 3226.30 3239.54 3220.30 -- 3 -- 24.14 8.67 -- -- 0.76
(2) Time added 3031.28 3053.36 3021.28 199.02*** 4 1 23.76 6.59 0.02 0.24 0.82
(3) Sleep added 2755.96 2829.14 2721.96 299.32*** 12 8 21.42 6.66 0.11 0.23 0.83
(4) Interactions added 2758.54 2857.54 2712.54 9.42 18 6 21.54 6.48 0.11 0.25 0.84
(5) Random effects added 2748.88 2865.10 2694.88 17.66** 22 4 18.15 5.64 0.25 0.35 0.86
Notes: AIC Akaike's Information Criterion; BIC Schwarz's Bayesian Criterion; -2LL, -2 log likelihood; A-2LL change in -2LL1 relative to
preceding model; s2b unexplained intercept-related (between subjects) variance; s2w = unexplained residual-related (within subjects) variance; rzb
Sbetween-subjects pseudo R-squared, an estimate of the amount of between subjects variance (estimated from null model) explained by fixed and
random predictors; r"w within-subjects pseudo R-squared, an estimate of the amount of within subjects variance (estimated from null model)
explained by fixed and random predictors; rit total pseudo R-squared, an estimate of total variance explained (estimated from amount of
agreement between predicted values and actual values). *** Deviance is significant at the 0.001 level. ** Deviance is significant at the 0.01 level.
* Deviance is significant at the 0.05 level.










Table 5-8. Sleep variables predicting reasoning
Fixed Effects

Predictor Variable B SE dSf t P
Within-person
Day 0.37 0.04 46.87 8.53 <0.001
Sleep Onset Latencys 0.01 0.01 494.50 1.94 0.05
Wake After Sleep Onsets 0.003 0.005 488.02 0.66 0.51
Terminal Wakefulnesss -0.003 0.01 490.64 -0.46 0.65
Sleep Onset Latencys o -0.004 0.01 480.81 -0.56 0.57
Wake After Sleep Onseto 0.01 0.02 459.99 0.69 0.49
Terminal Wakefulnesso 0.01 0.01 475.58 1.11 0.27
Between-person
Sleep Onset Latencys 0.002 0.03 47.48 0.08 0.94
Wake After Sleep Onsets -0.003 0.02 46.43 -0.14 0.89
Terminal Wakefulnesss -0.03 0.03 47.29 -0.90 0.37
Sleep Onset Latencys o -0.04 0.05 46.77 -0.69 0.50
Wake After Sleep Onseto -0.09 0.06 46.29 -1.56 0.13
Terminal Wakefulnesso 0.15 0.07 45.43 2.03 0.05
Interactions
Level 1 *Level 2 SOLs -0.00018 0.00007 482.38 -2.42 0.02
Level 1 *Level 2 WASOs -0.00002 0.00004 471.81 -0.54 0.59
Level 1 *Level 2 TWAKs 0.00018 0.00011 499.91 1.61 0.11
Level 1 Level 2 SOL o 0.00001 0.00013 394.80 0.09 0.93
Level 1 Level 2 WASO o -0.00033 0.00040 455.47 -0.82 0.41
Level 1 *Level 2 TWAK o -0.00026 0.00039 457.58 -0.67 0.51
Random Effects
Covariance parameter estimate B SE Z P
Within-person
Day 0.04 0.02 2.73 0.01
Sleep Onset Latencys 0.00 0.00 0.00 1.00
Level 1 *Level 2 SOLs 0.00 0.00 0.00 1.00
Level 1 *Level 2 TWAKs 0.00 0.00 0.00 1.00
Within Pseudo R2 0.35
Between Pseudo R2 0.25
Total Pseudo R" 0.86
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.












Table 5-9. Steps taken in building the Symbol Digit Multilevel Model
Symbol Digit
Models AIC BIC -2LL A-2LL dfAf s2b S2w ~ b wt
(1) Null 4498.87 4511.94 4492.87 -- 3 -- 68.42 121.52 --- 0.39
(2) Time added 4471.83 4493.62 4461.83 31.04*** 4 1 63.91 121.97 0.07 0.00 0.41
(3) Sleep added 4021.51 4093.63 3987.51 474.32*** 12 8 52.18 123.00 0.24 -0.01 0.43
(4) Interactions added 4018.75 4116.32 3972.75 14.76* 18 6 52.50 120.28 0.23 0.01 0.44
(5) Random effects added 4026.59 4141.13 3972.59 0.16 22 4 52.41 120.03 0.23 0.01 0.44
Notes: AIC Akaike's Information Criterion; BIC Schwarz's Bayesian Criterion; -2LL -2 log likelihood; A-2LL change in -2LL relative to
preceding model; s2b unexplained intercept-related (between subjects) variance; s2w = unexplained residual-related (within subjects) variance; r2b
Sbetween-subjects pseudo R-squared, an estimate of the amount of between subjects variance (estimated from null model) explained by fixed and
random predictors; rZw within-subjects pseudo R-squared, an estimate of the amount of within subjects variance (estimated from null model)
explained by fixed and random predictors; rit total pseudo R-squared, an estimate of total variance explained (estimated from amount of
agreement between predicted values and actual values). *** Deviance is significant at the 0.001 level. ** Deviance is significant at the 0.01 level.
* Deviance is significant at the 0.05 level.










Table 5-10. Sleep variables predicting processing speed
Fixed Effects

Predictor Variable B SE dSf t P
Within-person
Day 0.48 0.14 144.30 3.39 0.001
Sleep Onset Latencys -0.02 0.03 461.39 -0.56 0.58
Wake After Sleep Onsets -0.02 0.02 459.24 -1.15 0.25
Terminal Wakefulnesss -0.04 0.03 446.96 -1.58 0.12
Sleep Onset Latencys o 0.03 0.03 442.21 0.82 0.41
Wake After Sleep Onseto 0.14 0.08 421.86 1.82 0.07
Terminal Wakefulnesso -0.11 0.06 95.13 -1.91 0.06
Between-person
Sleep Onset Latencys 0.11 0.06 43.77 1.83 0.07
Wake After Sleep Onsets -0.08 0.04 41.47 -2.08 0.04
Terminal Wakefulnesss 0.07 0.06 43.36 1.14 0.26
Sleep Onset Latencys o -0.07 0.10 44.58 -0.71 0.48
Wake After Sleep Onseto 0.05 0.10 41.94 0.49 0.62
Terminal Wakefulnesso 0.09 0.14 42.52 0.69 0.49
Interactions
Level 1 *Level 2 SOLs 0.0002 0.0003 458.90 0.52 0.61
Level 1 *Level 2 WASOs 0.0002 0.0002 430.17 0.98 0.33
Level 1 *Level 2 TWAKs 0.0010 0.0005 465.84 2.13 0.03
Level 1 Level 2 SOL o -0.0005 0.0006 380.48 -0.90 0.37
Level 1 Level 2 WASO o -0.0028 0.0018 390.40 -1.60 0.11
Level 1 *Level 2 TWAK o 0.0042 0.0019 22.62 2.26 0.03
Random Effects
Covariance parameter estimate B SE Z P
Within-person
Day 0.00 0.00 0.00 1.00
Terminal Wakefulnesso 0.00 0.00 0.00 1.00
Level 1 *Level 2 TWAKs 0.00 0.00 0.00 1.00
Level 1 *Level 2 TWAK o 0.000001 0.000002 0.35 0.72
Within Pseudo R2 0.01
Between Pseudo R2 0.23
Total Pseudo R" 0.44
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.









CHAPTER 6
DISCUSSION

This section has four main obj ectives: to review and interpret the results of each primary

aim of the study; to consider the maj or limitations of the study; to discuss the larger theoretical

and empirical implications; and to discuss future directions for the study of the sleep-cognition

relationship.

Review of Findings

Aim 1: To Determine the Amount of Variability in Sleep and Cognition Found Within
Older Adults with Insomnia Compared to the Amount of Variability between Older
Adults with Insomnia

Results from this analysis indicate that a substantial amount of variability in sleep and

cognition is found within older adults with insomnia after controlling for any effects of time.

Variability in the letter series test was the lowest of all the variables measured and still

considerable (48%). Every other variable assessed exhibited either equal amounts of within-

person variability as between-person variability or more (range = 85% 189%). Results suggest

that within-person variability in sleep and cognition in older adults with insomnia is large enough

to warrant further investigation. Results are congruent with the literature in both the sleep and

cognitive field demonstrating large amounts of within-person variability in their relative

constructs of study (Allaire & Marsiske, 2005; Coates et al., 1981; Edinger et al., 1991; Edinger

et al., 1997; Frankel et al., 1976; Hauri & Wisbey, 1992; Hultsch et al., 2000; Hultsch et al.,

2002; MacDonald et al., 2003; Nesselroade & Salthouse, 2004; Salthouse et al., 2006; Salthouse

& Berish, 2005 Vallieres et al., 2005)

Aim 2: To Determine How Variability is Associated Across Sleep and Cognitive Measures
in Older Adults with Insomnia

The first question this aim addressed was how intraindividual variability is related across

the various sleep and cognition domains. Results indicated that variability in one domain is









related to variability in other domains. It appears that variability may operate systematically

within-persons. Individuals who operate in a variable manner in one task also operate variably in

other tasks. These findings are inline with the bulk of previous studies of intraindividual

variability (Allaire & Marsiske, 2005; Hultsch et al., 2000; Nesselroade & Salthouse, 2004).

Interestingly, variability in WASOs was negatively, though non-significantly, correlated with

variability in cognition.

The second question this aim addressed was how level of performance is related across the

various sleep and cognition domains. Interestingly, results indicate that in general better sleep is

related to lower letter series performance and higher symbol digit scores. Although not

definitive, this pattern of findings supplies some evidence for the notion that the symbol digit test

may be more indicative of underlying brain functions than the letter series test.

The third question this aim addressed was how variability is related to level of

performance across the various sleep and cognition measures. Increased variability in both letter

series and symbol digit were related to increased overall level of sleep. This suggests that

individuals who had higher mean values on the sleep variables had more variable cognitive

performance. In light of the fact that all of the sleep parameters assessed are typically used as

indicators of poor sleep and assuming more variability in cognition is a bad thing, this finding

make sense. If variability is indicative of underlying neuronal dysfunction (Sliwinski, Hofer, &

Hall, 2003), this result lends support to the notion that sleep provides a restorative period for the

brain. Increased variability in sleep parameters was generally negatively associated with average

performance on the letter series task and positively associated with average symbol digit score.

Although seemingly contradictory in indications, these results do fit prior research findings

suggesting that variability may be adaptive (a good thing) or a sign of neural deterioration (a bad










thing) (Allaire & Marsiske, 2005). Increased variability in sleep was associated with poorer

overall sleep. Variable sleep is typically viewed as an indication of poor sleep, as suggested by

the findings that insomniacs tend to be more variable in their sleep patterns than normal 'good'

sleepers (Coates et al., 1981; Edinger et al., 1997). This set of results confirms that more variable

sleep is associated with poorer sleep. Increased variability in both reasoning and processing

speed were found to be associated with better reasoning and processing speed. These results add

to the current literature suggesting that variability may be adaptive in nature (Allaire & Marsiske,

2005; Li, Aggen, Nesselroade, & Baltes, 2001; Siegler, 1994).

Aim 3: To Determine the Between-Person (Mean-Level) and Within-Person (Day-to-Day-
Level) Association between Sleep and Cognition in Older Adults with Insomnia

An initial correlational analysis of predictor variables, both within and between-persons,

revealed that no two within-person predictors shared more than 2.5% of their variance and that

no two between-person predictors shared more than 55% of their variance.

MLM for reasoning

The intraclass correlation for letter series was 0.36 indicting that 64% of the variance in

letter series was within-persons and investigation of this within-person variability was warranted.

After controlling for time, within-person variations in SOLs significantly predicted letter

series performance. Results suggest that on days when an individual self-estimates taking longer

to fall asleep, they also have better reasoning performance. A significant positive between-person

association was found between TWAKo and letter series indicating that individuals who, on

average, tend to spend more time laying in bed in the morning also have better than average

reasoning. TWAK, in general, has been an understudied variable. Results that increased TWAK

may be beneficial to cognitive performance could indicate an attempt to compensate for a poor










night' s sleep that works. Yet, TWAK is largely considered maladaptive in sleep research. Thus,

additional research is needed before definitive conclusions can be drawn.

There was also a significant negative interaction between within-person and between-

person SOLs indicating that when an individual who usually has above average SOLs

experiences a day of increased SOLs they perform worse on the reasoning task. Though these

results may seem puzzling at first they are very much inline with current findings on the sleep-

cognition relationship.

The fact that when individuals experience above average SOLs and when individuals tend

to, on average, have more TWAKo they also perform better on the letter series could be evidence

for the hypothesized added effort such individuals are suspected of expending to offset the

natural decline following poor sleep (Drake et al., 2001). Furthermore, when individuals who

usually take a long time to fall asleep take even longer they consequently experience lower than

average reasoning. This could be interpreted as creating too much impairment for added effort to

compensate for, as in total sleep deprivation.

This model accounted for 3 5% of the within-person variance in letter series and 25% of the

between-person variance. In total, the model accounted for 86% of the variance in reasoning

performance.

MLM for processing speed

The intraclass correlation for letter series was 0.73 indicting that 27% of the variance in

letter series was within-persons and investigation of this within-person variability was warranted.

After controlling for time there were no significant within-person predictors of symbol

digit performance. A negative significant between-person association between WASOs and

symbol digit was found. This relationship indicates that individuals who, on average, spend more

self-reported wake time in the middle of the night also have lower than average processing










speeds. This finding is consistent with the sleep deprivation researching showing a negative

relationship between sleep loss and cognitive performance in older adults (Webb, 1985; Web &

Levy, 1982).

Two significant interactions were found. The within-person by between person interaction

for both TWAKs and TWAKo were positively related to symbol digit performance. These

interactions indicate that when an individual who usually has above average TWAK experiences

a day of increased TWAK they have better processing speeds. As stated above, TWAK is a

recently introduced variable into the sleep literature. These interactions suggest that getting much

more rest time in the morning may supply a type of restorative function for the prefrontal cortex

that was not achieved at night (Horne, 1988). Increasing TWAK may be an adaptive behavior to

the loss of nocturnal sleep in order to maintain optimal cognitive functioning. However, as stated

previously, TWAK is generally agreed upon to be a maladaptive sleep response and thus needs

further research before definitive conclusions are drawn.

This model accounted for only 1% of the within-person variance in symbol digit indicating

the need for better within-person predictors. However, the model accounted for 23% of the

between-person variance and 44% of the total variance in processing speed.

Study Limitations

There are several important study limitations that need to be addressed. While the data

presented here were well suited to answer the questions asked; this is a secondary data analysis.

Therefore, it is important to recognize that the data were not collected to answer the questions of

within-person variability in sleep and cognition over 14 days in older adults with insomnia.

However, it also must be recognized that the data were collected for a closely related purpose

and as such do fit the current uses extremely well.









The lack of a "normal" sleeping control group restricts the interpretation of results. The

presence of such a comparison would have allowed statements about the differing magnitude,

differing associations, and differing daily coupling of sleep and cognition. As is, the current

study is confined to addressing these important questions in older insomniacs only.

Generalizability of results may be further hindered by the sample characteristics of the study.

Specifically, study participants were above-average educated and mostly Caucasian.

Lastly, the use of self-administered daily assessments and the resulting loss of

experimental control must be acknowledged. Participants were thoroughly instructed in the

procedures of all material and were provided with all necessary material and then instructed to

complete everything once daily. Although the possibility of data integrity problems definitely

exists, we believe this is highly unlikely for several reasons. A very similar protocol was used in

a previous study of intraindividual variability successfully (Allaire & Marsiske, 2005). Self-

assessments were also monitored weekly to decrease the probability of incorrect administration.

And, all workbooks contained integrity statements that were all signed and returned.

Theoretical and Empirical Implications

With these limitations in mind, we feel this study has several important implications for

sleep researchers, cognitive aging researchers, and "real world" applications. First, this research

reveals the importance of within-person variations in sleep. Intraindividual variability has yet to

become considered a salient topic in sleep research. Instead many researchers are focusing

exclusively on interindividual variability at the expense of intraindividual variability. Results of

the current study indicate that within-person variations in sleep are not "noise"' or measurement

error and suggest the need to systematically study how individuals' sleep varies from day-to-day.

Specifically, the results of Aim 3 illustrate that disregarding within-person variability as

measurement error would lead to decreased precision in predicting cognition with sleep.









It is also hoped that the results of this study will inspire more sleep researchers to become

familiar with multilevel modeling and its application for studying within-person variability.

Several researchers have previously employed multilevel modeling but failed to decompose

predictors into within and between components (Monk et al., 2006). By doing this these

researchers may have missed possibly intriguing results about within-person processes.

Furthermore, several researchers have commented on the possibility that the inconsistencies in

the results of sleep deprivation studies are due to stable interindividual differences in

susceptibility to sleep loss (Olofsen et al., 2004; Van Dongen et al., 2004). It is our contention

that through the use of MLM and testing random effects such hypotheses can be tested.

This study adds to the sleep research literature an investigation of the sleep-cognition

relationship that is ecologically valid. While many studies have deprived people of their sleep

and then tested their cognitive functioning (Harrison & Horne, 2000; Pilcher & Huffcutt, 1996)

very few have attempted to capture this relationship in an already poor sleeping sample.

Furthermore, those studies that did assess individuals in their natural setting settled for relatively

brief cognitive measures instead of commonly used, domain specific assessment (Blackwell et

al., 2006; Cricco et al., 2001; Tworoger et al., 2006). Aim 3 also points for the need of increased

research into the utility of TWAK as a sleep variable. TWAK is traditionally viewed as an un-

recommended response to poor nocturnal sleep and as a contributor to subsequent nights of poor

sleep. Our results suggest that TWAK may be beneficial to cognitive performance in older adults

with insomnia. We suggest future research examine this relationship in more detail. In this way,

the current study adds depth to the current state of sleep research examining the sleep-cognition

link.









Implications for the cognitive aging field are straightforward; sleep is associated with

cognitive performance in older adults. Given the known changes in sleep that occurs with

increased age (Morgan, 2000) and the known relationship between sleep and cognition (Harrison

& Horne, 2000; Pilcher & Huffcutt, 1996) it is essential that sleep be more readily studied within

the cognitive aging field. Is it coincidence that executive functioning, which is regulated by the

prefrontal cortex, has been found to exhibit high levels of variability (Salthouse et al., 2006) and

that the prefrontal cortex is the most sensitive region of the brain to sleep loss (Durmer &

Dinges, 2005) and that sleep exhibits much variability in "normal" sleepers and older adults with

insomnia (Coates et al., 1981; Edinger et al., 1991; Edinger et al., 1997; Frankel et al., 1976;

Hauri & Wisbey, 1992; Vallieres et al., 2005; McCrae et al., 2003; McCrae et al., 2005)?

Implications of the current study for real world applicability are many. First, a reliable

relationship was illustrated between sleep and cognitive functioning. It is therefore important for

indivi duals doing cognitive/neurop sychol ogi cal assessments of older adults to also asse ss sleep.

Given the association between sleep loss and accidents, auto and work related, (Benca, 2001;

Hublin & Partinen, 2002; Katz & McHorney, 2002; Neubauer, 2004; Roth & Roehrs, 2003) and

the daily associations between sleep and perceptual speed found in the current study, it is

imperative to warn older drivers of the potential negative consequences of one night' s bad sleep.

There is a general cognitive decline experienced with increasing age (Craik & Byrd, 1982;

Hasher & Zacks, 1988; Lindenberger & Baltes, 1994; Salthouse, 1996). There is also a general

worsening of sleep associated with advanced age (Morgan, 2000). If these two phenomena are

causally related then a possible intervention to help slow one possible factor in age-related

cognitive decline could be treatments aimed at improving sleep. Future research should

investigate the longitudinal effects on cognitive performance of improved sleep.









Future Directions

Possible follow-up studies to the current proj ect could proceed in many different

directions. Given the possible relationship between REM sleep and memory consolidation (Gais

& Born, 2004; Wagner et al., 2001) and facilitation (Walker & Stickgold, 2004) a potential line

of research would be to have individuals wear portable polysomnographies equipped with

electroencephalograms nightly and measure word recall daily to determine the daily association

between amount of REM sleep obtained and memory performance.

Similar studies as the current one could be conducted that included normal controls for

comparison purposes. Further, future studies should examine more occasions, in a larger more

diverse sample, and should assess more cognitive domains using more precise tools, such as

computers, to allow for an accurate recording of individual reaction times. Such innovation

would allow for the examination of not only occasion-to-occasion variability but also within

occasion variability.

Another possible vein of research would be the detection of other covariates that could be

incorporated into the MLM to aid in the prediction of cognitive performance. Possible covariates

might include: affect, physical activity, life stressors, and pain.









APPENDIX A
SLEEP DIARY

Please answer the following questionnaire WHEN YOU AWAKE IN THE MORNING. Enter
yesterday's day and date and provide the information to describe your sleep the night before.
Definitions explaining each line of the questionnaire are given below.

Table A-1. Example of a Sleep, Diar
yesterday's day TUES
yesterday's date 10/14/9


day 1 day 2 day 3 day 4 day 5 day 6 day 7

1. NAP (yesterday) 70

2. BEDTIMVE (last 10:55

night)

3. TIME TO FALL 65

ASLEEP

4. # AWAKENINGS 4

5. WAKE TIME 110
(middle of night)
6. FINAL WAKE-UP 6:05

7. OUT OF BED 7: 10

8. QUALITY RATING* 2

9. BEDTIME Halcion

MEDICATION 0.25 mg

(include amount & 10:40 pm

time)

*Pick one number below to indicate your ove -all QUALITY 10 TING or satisfac ion with your
sleep.1. very poor, 2. poor, 3. fair, 4. good, 5. excel ent










APPENDIX B
LETTER SERIES TEST

Circle the one letter from the answer choices on the ri ht that comes next in
the series of letters on the left.
Answer Choices

1. oo pq qr ss t uu oq tu v


2. op qo pq rs tr st u vw u vwx y

3. o s o t p s p t q s q t r sr p q r s t


4. f g h p f g h q f g h rf g h f q rs t


5. no pi jk qr si jk t uv wx yi j

6. u vwuv xu vy u vz u vwy z


7.d ep fg ph ip j k plmp d el1mn


8. qp oq po qp o qp op qr s


9. pa pq ap qr ap q rs rs tu a

10.st st pq u vuv p qwx p qwx y


11.o pp qq qr rr rs s ss rs tu v

12. ooqq ss u uww v wxy z

13.r st rs tu rs t uv rs tu v


14. p qq rs tt u vwwx yz v wxy z








Answer Choices
opqrs

q r svw

opqrs

ij ko p

st uv w


mnopq

i jkpq

pqrst

qrstu

nopfg

qrstu

pqrst

mn op q

mnopq

pqrst

nopqr


15.i j i klkmnmopoqr

16.ij k vwl mnyvwo pq v w

17.i j j j klllmnn no p p

18.p onmlkj

19.i kmoqsu


20.i j j kllmnn o p

21.i pj qk ri pj qk ri p j

22.h i k no qr


23.i j k ilmnlo pq or s t

24.ia j b k cldmen f o

25.a cf h kmpr

26.ii j j kllmmnoop

27.i ij i jkklklmmn

28.n nn nn oo oo pp p q


29.i jk k j ilmn nmlo p q

30.ij k j klmn mn op q p


















C >r -C-=-Ir -l>)+ r -1-CIt>
4 1 2 9 4 5 2 7 11 6


1- = )> -1 +rl > -= Ir F + --



+ 1- ) r l I-: -1 > .. C )I -= )



--r -=-) + I ) -) > =-+ C -



r I-C > -1+> C 1) 1 + --


APPENDIX C
SYMBOL DIGIT TEST
On the next page are boxes to be filled in with the matching digit, like you see in the key
on the top of the page. Go in order, without skipping any boxes.

KEY


1r 2 3t 4 56 78







-1 + CI -= F> > -C+ -1 -' ) r


)r>+1I --C-=-)-r>-)C I


> C -1 ) -=-r C > + IC 1- -1 r










APPENDIX D
CORRELATIONS OF PREDICTOR VARIABLES NOT CONTROLLING FOR TIMVE

Table D-1. Correlations of within-person, with time, predictors. (N=48, 14 Occasions)
1 2 3 4 5 6 7 8
1. Letter Series 1.00
2. Symbol Digit 0.16** 1.00
3. Sleep Onset Latency 0.01 0.01 1.00
4. Wake After Sleep Onset 0.03 0.02 -0.05 1.00
5. Terminal Wakefulness 0.05 0.03 0.10** 0.05 1.00
6. Sleep Onset Latency -0.02 0.00 0.06 0.01 0.01 1.00
7. Wake After Sleep Onset 0.03 0.05 0.08 0.14** 0.11** -0.10* 1.00
8. Terminal Wakefulness 0.06 0.04 -0.05 -0.05 0.15** 0.15** -0.12** 1.00
Note: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the
0.05 level (2-tailed). Variables with a subscript 's' indicate they were subjectively measured by
Sleep Diary. Variables with a subscript 'o' indicate they were objectively measured by
Actigraphy.



Table D-2. Correlations of between-person, with time, predictors. (N=48)
1 2 3 4 5 6 7 8


1. Letter Series 1.00
2. Symbol Digit 0.23 1.00
3. Sleep Onset Latency -0.14 0.30 1.00
4. Wake After Sleep Onset -0.06 -0.07 0.48** 1.00
5. Terminal Wakefulness -0.15 0.25 0.45* 0.48** 1.00
6. Sleep Onset Latency -0.11 0.21 0.74** 0.11 0.29 1.00
7. Wake After Sleep Onset -0.18 0.13 0.16 0.26 0.18 0.08 1.00
8. Terminal Wakefulness 0.14 0.26 0.41 0.34 0.41 0.45* 0.50** 1.00
Note: ** Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the
0.05 level (2-tailed). Variables with a subscript 's' indicate they were subjectively measured by
Sleep Diary. Variables with a subscript 'o' indicate they were subjectively measured by
Actigraphy.













Table E-1. Reasoning MLM Step 2: Adding Time
Fixed Effects

Predictor Variable B SE dSf t P
Within-person
Day 0.38 0.03 148.29 12.40 <0.00
Between-person

Interactions -- -- -- -- --

Random Effects
Covariance parameter estimate B SE Z P
Within-person

Within Pseudo R 0.24
Between Pseudo R2 0.02
Total Pseudo R" 0.82
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.


APPENDIX E
REASONING MODELS AT EACH STEP OF MLM BUILDING










Table E-2. Reasoning MLM Step 3: Adding Sleep
Fixed Effects

Predictor Variable B SE dSf t P
Within-person
Day 0.37 0.03 140.29 11.42 <0.00
Sleep Onset Latencys 0.00 0.00 494.24 0.05 0.96
Wake After Sleep Onsets 0.00 0.00 481.36 0.19 0.85
Terminal Wakefulnesss 0.01 0.00 498.63 1.73 0.08
Sleep Onset Latencys o 0.00 0.00 439.32 -0.69 0.49
Wake After Sleep Onseto 0.00 0.01 489.10 -0.06 0.95
Terminal Wakefulnesso 0.01 0.00 477.54 1.12 0.26
Between-person
Sleep Onset Latencys -0.01 0.03 45.74 -0.29 0.77
Wake After Sleep Onsets 0.01 0.02 45.37 0.28 0.78
Terminal Wakefulnesss -0.03 0.03 45.69 -0.92 0.36
Sleep Onset Latencys o -0.03 0.06 45.87 -0.56 0.58
Wake After Sleep Onseto -0.09 0.06 45.39 -1.58 0.12
Terminal Wakefulnesso 0.13 0.08 45.48 1.73 0.09
Interactions

Random Effects
Covariance parameter estimate B SE Z P
Within-person

Within Pseudo R2 0.23
Between Pseudo R2 0.11


Total Pseudo R


0.83


Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.





Predictor Variable B SE dSf t P
Within-person
Day 0.37 0.03 141.86 11.57 <0.00
Sleep Onset Latencys 0.01 0.01 497.63 1.97 0.05
Wake After Sleep Onsets 0.00 0.00 488.99 0.46 0.64
Terminal Wakefulnesss 0.00 0.01 484.46 -0.53 0.60
Sleep Onset Latencys o 0.00 0.01 474.18 -0.45 0.65
Wake After Sleep Onseto 0.01 0.02 452.80 0.53 0.59
Terminal Wakefulnesso 0.01 0.01 462.95 0.89 0.38
Between-person
Sleep Onset Latencys -0.01 0.03 45.72 -0.29 0.77
Wake After Sleep Onsets 0.01 0.02 45.35 0.24 0.81
Terminal Wakefulnesss -0.03 0.03 45.65 -0.90 0.37
Sleep Onset Latencys o -0.03 0.06 45.84 -0.57 0.57
Wake After Sleep Onseto -0.09 0.06 45.38 -1.57 0.12
Terminal Wakefulnesso 0.13 0.08 45.45 1.73 0.09
Interactions
Level 1 *Level 2 SOLs 0.00 0.00 496.60 -2.40 0.02
Level 1 *Level 2 WASOs 0.00 0.00 459.20 -0.46 0.65
Level 1 *Level 2 TWAKs 0.00 0.00 497.89 1.99 0.05
Level 1 Level 2 SOL o 0.00 0.00 397.25 -0.01 0.99
Level 1 Level 2 WASO o 0.00 0.00 453.78 -0.69 0.49
Level 1 *Level 2 TWAK o 0.00 0.00 440.46 -0.52 0.61
Random Effects
Covariance parameter estimate B SE Z P
Within-person

Within Pseudo R" 0.25
Between Pseudo R2 0.11


Table E-3. Reasoning MLM Step 4: Adding Interactions
Fixed Effects


Total Pseudo R2 0.84
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.













Table F-1. Processing Speed MLM Step 2: Adding Time
Fixed Effects

Predictor Variable B SE dSf t P
Within-person
Day 0.48 0.13 160.44 3.59 <0.00
Between-person

Interactions

Random Effects
Covariance parameter estimate B SE Z P
Within-person

Within Pseudo R 0.00
Between Pseudo R2 0.07
Total Pseudo R" 0.41
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.


APPENDIX F
PROCESSING SPEED MODELS AT EACH STEP OF MLM BUILDING










Table F-2. Processing Speed MLM Step 3: Adding Sleep
Fixed Effects

Predictor Variable B SE dSf t P
Within-person
Day 0.47 0.14 148.90 3.36 <0.00
Sleep Onset Latencys 0.00 0.01 470.82 -0.27 0.79
Wake After Sleep Onsets 0.00 0.01 456.11 -0.25 0.80
Terminal Wakefulnesss 0.01 0.01 469.90 0.49 0.62
Sleep Onset Latencys o 0.00 0.02 414.49 0.05 0.96
Wake After Sleep Onseto 0.02 0.03 468.20 0.72 0.47
Terminal Wakefulnesso 0.02 0.02 451.04 0.98 0.33
Between-person
Sleep Onset Latencys 0.10 0.06 43.65 1.80 0.08
Wake After Sleep Onsets -0.08 0.04 41.32 -2.05 0.05
Terminal Wakefulnesss 0.07 0.06 43.26 1.14 0.26
Sleep Onset Latencys o -0.07 0.10 44.48 -0.65 0.52
Wake After Sleep Onseto 0.05 0.10 41.85 0.52 0.61
Terminal Wakefulnesso 0.10 0.14 42.42 0.71 0.48
Interactions

Random Effects
Covariance parameter estimate B SE Z P
Within-person

Within Pseudo R2 -0.01
Between Pseudo R2 0.24
Total Pseudo R" 0.43
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.





Predictor Variable B SE dSf t P
Within-person
Day 0.48 0.14 146.19 3.36 <0.00
Sleep Onset Latencys -0.02 0.03 465.87 -0.56 0.58
Wake After Sleep Onsets -0.02 0.02 459.51 -1.10 0.27
Terminal Wakefulnesss -0.04 0.03 455.71 -1.57 0.12
Sleep Onset Latencys o 0.03 0.03 444.73 0.80 0.43
Wake After Sleep Onseto 0.15 0.08 437.49 1.89 0.06
Terminal Wakefulnesso -0.11 0.06 428.38 -2.03 0.04
Between-person
Sleep Onset Latencys 0.11 0.06 43.76 1.84 0.07
Wake After Sleep Onsets -0.08 0.04 41.46 -2.08 0.04
Terminal Wakefulnesss 0.07 0.06 43.35 1.14 0.26
Sleep Onset Latencys o -0.07 0.10 44.57 -0.71 0.48
Wake After Sleep Onseto 0.05 0.10 41.94 0.49 0.62
Terminal Wakefulnesso 0.09 0.14 42.51 0.69 0.49
Interactions
Level 1 *Level 2 SOLs 0.00 0.00 465.48 0.51 0.61
Level 1 *Level 2 WASOs 0.00 0.00 433.48 0.96 0.34
Level 1 *Level 2 TWAKs 0.00 0.00 467.09 2.14 0.03
Level 1 Level 2 SOL o 0.00 0.00 382.83 -0.87 0.38
Level 1 Level 2 WASO o 0.00 0.00 429.51 -1.70 0.09
Level 1 *Level 2 TWAK o 0.00 0.00 410.50 2.51 0.01
Random Effects
Covariance parameter estimate B SE Z P
Within-person

Within Pseudo R" 0.01
Between Pseudo R2 0.23


Table F-3. Processing Speed MLM Step 4: Adding Interactions
Fixed Effects


Total Pseudo R2 0.44
Notes: Variables with a subscript 's' indicate they were subj ectively measured by Sleep Diary.
Variables with a subscript 'o' indicate they were obj ectively measured by Actigraphy.











LIST OF REFERENCES


Allaire, J. C., & Marsiske, M. (2005). Intraindividual variability may not always indicate
vulnerability in elders' cognitive performance. Psychology and Aging, 20, 390-401.

American Academy of Sleep Medicine. (2005). The International Classification of Sleep
Disorders: Diagnostic and Coding Manual, 2nd Edition. Westchester, IL: American
Academy of Sleep Medicine.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental
disorders. Washington, DC: Author.

Bastein, C. H., Fortier-Brochu, E., Rioux, I., LeBlanc, M., Daley, M., & Morin, C. M. (2003).
Cognitive performance and sleep quality in the elderly suffering from chronic insomnia:
Relationship between obj ective and subj ective measures. Journal ofPsychosomatic
Research, 54, 39-49.

Beck, A. T., Brown, G., & Steer, R. A. (1996). Beck Depression Inventory H manual. San
Antonio, TX: The Psychological Corporation.

Belenky, G., Wesenten, N. J., Thorne, D. R., Thomas, M. L., Sing, H C., Redmond, D P., Russo,
M. C., & Balkin, T. J. (2003). Patterns of performance degradation and restoration during
sleep restriction and subsequent recovery: a sleep dose-response study. Journal of Sleep
Research, 12, 1-12.

Benca, R. M. (2002). Consequences of insomnia and its therapies. Journal of Clinical
Psychiatry, 62, 33-38.

Blackwell, T., Yaffe, K., Ancoli-Isreal, S., Schneider, J. L., Cauley, J. A., Hiller, T. A., Fink, H.
A., & Stone, K. L. (2006). Poor sleep is associated with impaired cognitive function in
older women: The study of osteoporotic fractures. Journal of Gerontology, 61, 405-410.

Bliwise, D. (1993). Sleep in normal aging and dementia. Sleep, 16, 40-81.

Bosselli, M., Parrino, L., Smerieri, A., & Terzano, M. G. (1998). Effects of age on the EEG
arousals in normal sleep. Sleep, 21, 351-357.

Bryk, A. S. & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data
analysis methods. Thousand Oaks: Sage.

Burton, C. L., Hultsch, D. F., Strauss, E., & Hunter, M. A. (2002). Intraindividual variability in
physical and emotional functioning: Comparison of adults with traumatic brain injury and
healthy adults. The Clinical Neuropsychologist, 16, 264-279.

Butler, A. C., Hokanson, J. E., & Flynn, H. A., (1994). A comparison of self-esteem liability and
low trait self-esteem as vulnerability factors for depression. Journal of Personality and
Social Psychology, 66, 166-177.










Craik, F. I. M., & Byrd, M. (1982). Aging and cognitive deficits: The role of attentional
resources. In F. I. M. Craik & S. Trehub (Eds.), Aging and' cognitive processes. New York:
Plenum Press.

Coates, T. J., George, J. M., Killen, J. D., Marchini, E., Hamilton, S. and Thorensen, C. E.
(1981). First night effects in good sleepers and sleep-maintenance insomniacs when
recorder from home. Sleep, 4, 293-8.

Crenshaw, M. C., & Edinger, J. D. (1999). Slow-wave sleep and waking cognitive performance
among older adults with and without insomnia complaints. Physiology & Behavior, 66,
485-492.

Cricco, M., Simonsick, E. M., & Foley, D. J. (2001). The impact of insomnia on cognitive
functioning in older adults. Journal of the American Geriatrics Society, 49, 1185-1189.

Dinges, D. F., Pack, F., Williams, K., Gillen, N. A., Powell, J W., Ott, G E., Aptowicz, C., &
Pack, A. I. (1997). Cumulative sleepiness, mood disturbance, and psychomotor vigilance
performance decrements during a week of sleep restricted to 4-5 hours per night. Sleep, 20,
267-277.

Drake, C. L., Roehrs, T. A., Burduvali, E., Bonahoom, A., Rosekind, M., & Roth, T. (2001).
Effects of rapid versus slow accumulation of eight hours of sleep loss. Psychophysiology,
38, 979-987.

Drummond, S. P. A., Brown, G. A., Gillin, J. C., Stricker, J. L., Wong, E. C., & Buxton, R. B.
(2000). Altered brain response to verbal learning following sleep deprivation. Nature, 403,
655-657.

Durmer, J. S., & Dinges, D. F. (2005). Neurocognitive consequences of sleep deprivation.
Seminars in Neurology, 25, 117-129.

Edinger, J. D., Fins, A. I., Sullivan, R. J., Jr., Marsh, G. R., Dailey, D. S., Hope, T. V., Young,
M., Shaw, E., Carlson, D. and Vasilas, D. (1997). Sleep in the laboratory and sleep at
home: Comparisons of older insomniacs and normal sleepers. Sleep, 20, 1 119-26.

Edinger, J. D., Marsh, G. R., Mccall, W. V., Erwin, C. W. and Lininger, A. W. (1991). Sleep
variability across consecutive nights of home monitoring in older mixed DIMS patients.
Sleep, 14, 13-17.

Eizenman, D. R., Nesselroade, J. R., Featherman, D. L., & Rowe, J. W. (1997). Intraindividual
variability in perceived control in an older sample: The McArthur Successful Aging
Studies. Psychology and Aging, 12, 489-502.

Espie, C. A. (1991) The psychological treatment of insomnia. Chichesster, UK: J. Wiley and
Sons.










Floyd, J. A., Medler, S. M., Ager, J. W., & Janisse, J. J. (2000). Age-related changes in the
initiation and maintenance of sleep: A meta-analysis. Research in Nursing and Health, 23,
106-117.

Folstein, M. F., Folstein, S. E. & McHugh, P. R. (1975). "Mini Mental State": A practical
method for grading the cognitive states of patients for the clinician. Journal of Psychiatric
Research, 12, 189-198.

Frankel, B. L., Coursey, R. D., Buchbinder, R. and Snyder, F. (1976). Recorded and reported
sleep in chronic primary insomnia. Archives of General Psychiatry, 33, 6 15-23.

Gais, S., & Born, J. (2004). Low acetycholine during slow-wave sleep is critical for declarative
memory consolidation. Proceedings of the National Academy of Science, 10O1, 2 140-2 144.

Harrison, Y., & Horne, J. A. (2000). The impact of sleep deprivation on decision making: A
review. Journal ofExperintental Psychology, 6, 236-249.

Hasher, L., & Zacks, R. T., (1988). Working memory, comprehension and aging: A review and a
new view. In G. H. Bower (ed.), 7Jhe psychology of learning and motivation. San Diego:
Academic Press.

Hauri, P. J. and Wisbey, J. (1992). Actigraphy and insomnia: A closer look. Sleep, 15, 293-301.

Hobson, J. A., & Pace-Schott, E. F. (2002). The cognitive neuroscience of sleep: Neuronal
systems, consciousness and learning. Nature Reviews, 3, 679-693.

Horne, J. A. (1988). Sleep deprivation and divergent thinking ability. Sleep, 11, 528-536.

Horne, J. A. (1993). Human sleep, sleep deprivation and behaviour: Implications for the
prefrontal cortex and psychiatric disorder. British Journal ofPsychiatry, 162, 413-419.

Hublin, C. G. M., & Partinen, M. M. (2002). The extent and impact of insomnia as a public
health problem. Journal of Clinical Psychiatry, 4, 8- 12.

Hultsch, D. F., MacDonald, S. W. S., & Dixon, R. A. (2002). Variability in reaction time
performance of younger and older adults. Journal ofGerontology, 57, 101-1 15.

Hultsch, D. F., MacDonald, S. W. S., Hunter, M. A., Levy-Bencheton, J., & Strauss, E. (2000).
Intraindividual variability in cognitive performance in older adults: Comparison of adults
with mild dementia, adults with arthritis, and healthy adults. Neuropsychology, 14, 588-
598.

Katz D. A., & McHorney C. A. (2002). The relationship between insomnia and health-related
quality of life in patients with chronic illness. Journal ofFa~nily Practice, 5 1, 229-23 5.

Kreft, I. G. G., de Leeuw, J., & Aiken, L. S. (1995). The effect of different forms of centering in
hierarchical linear models. Midltivariate Behavioral Research, 30, 1-21.










Krueger, G. P. (1989). Sustained work, fatigue, sleep loss and performance: a review of the
issues. Work Stress, 3, 129-141.

Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). New York: Oxford University
Press.

Lezak, M. D. (2004). Neuropsychological assessment (4th ed.). New York: Oxford University
Press.

Li, S., Aggen, S. H., Nessleroade, J. R., & Baltes, P. B. (2001). Short-term fluctuations in elderly
people's sensorimotor functioning predict text and spatial memory performance: The
MacArthur Successful Aging Studies. Gerontology, 47, 100-116.

Lichstein, K. L., Riedel, B. W., & Means, M. K. (1999). Psychological treatment of late-life
insomnia. In R. Schulz, G. Maddox & M. P. Lawton (Eds.), Annual review of gerontology
and geriatrics: Vol. 18. Focus on interventions research aI ithr older adults (pp. 74-110).
New York: Springer.

Lindenburger, U., & Baltes, P. B. (1994). Sensory functioning and intelligence in old age: A
strong connection. Psychology and Aging, 9, 339-355.

MacDonald, S. W. S., Hultsch, D. F., & Dixon, R. A. (2003). Performance variability is related
to change in cognition: Evidence from the Victoria Longitudinal Study. Psychology and
Aging, 18, 510-523.

McCoy, K. J. M. (2004). Understanding the transition from normal cognitive aging to mild
cognitive impairment: Comparing the intraindividual variability in cognitive function.
Dissertation Abstracts International, University of Florida.

McCrae, C. S., Rowe, M. A., Tierney, C. G., Dautovich, N. D., DeFinis, A. L., & McNamara, J.
P. H. (2005). Sleep complaints, subjective and objective sleep patterns, health,
psychological adjustment, and daytime functioning in community-dwelling older adults.
Journal of Gerontology, 60, 1 82- 189.

McCrea, C. S., Wilson, N. M., Lichstein, K. L., Durrence, H. H., Taylor, D. J., Bush, A. J., &
Riedel, B. W. (2003). 'Young old' and 'old old' poor sleepers with and without insomnia
complaints. Journal ofPsychosomatic Research, 54, 11-19.

Mini Mitter Co., I. (2001). Actiwatch 16/Actiwatch 64/Actiwatch-L/Actiwatch-Score instruction
manual: Mini Mitter Co., Inc.

Monk, T. H., Thompson, W. K., Buysee, D. J., Hall, M., Nofzinger, E. A., & Reynolds, C. F.
(2006). Sleep in healthy seniors: A diary study of the relation between bedtime and the
amount of sleep obtained. Journal of Sleep Research, 15, 256-260.

Morgan, K. (1996). Mental health factors in late-life insomnia. Reviews in Clinical Gerontology,
6, 75-83.










Morgan, K. (2000). Sleep and Aging. In K. Lichstein & C. Morin (Eds.), Treatment of lte-life
insomnia. Thousand Oaks, CA: Sage Publications.

Muzur, A., Pace-Schott, E. F., & Hobson, J. A. (2002). The prefrontal cortex in sleep. Trends in
Cognitive Sciences, 6, 475-48 1.

Nesselroade, J. R., & Salthouse, T. A. (2004). Methodological and theoretical implications of
intraindividual variability in perceptual-motor performance. Journal ofGerontology, 59,
49-55.

Neubauer, D. N. (2004). Chronic insomnia: current issues. Clinical Cornerstone, 6, S17-S22.

Neupert, S. D., Almeida, D. M., Mroczek, D. K., & Spiro, A. (2006). Daily stressors and
memory failures in a naturalistic setting: Findings from the VA Normative Aging Study.
Psychology and Aging, 21, 424-429.

Newman, A. B., Enright, P. L., Manolio, T. A., Haponik, E. F., & Wahl, P. W. (1997). Sleep
disturbance, psychological correlates, and cardiovascular disease in 5201 older adults: The
cardiovascular health study. Journal of the American Geriatrics Society, 45, 1-7.

Ohayon, M. (1996). Epidemiological study on insomnia in the general population. Sleep, 19, S7-
S15.

Ohayon, M. (2002). Methodology of a study on insomia in the general population. Enciphale,
28, 217-226.

Olofsen, E., Dinges, D. F., & Van Dongen, H. P. A. (2004). Nonlinear mixed-effects modeling:
Individualization and prediction. Aviation, Space, and Environmental M~edicine, 75, 13 4-
140.

Pallesen, S., Nordhus, I. H. and Kvale, G. (1998). Nonpharmacological interventions for
insomnia in older adults: A meta-analysis of treatment efficacy. Psychotherapy, 3 5, 472-
482.

Pilcher, J. J., & Huffcut, A. J. (1996). Effects of sleep deprivation on performance: A meta-
analysis, Sleep, 19, 318-326.

Prinz, P. N., Peskind, E. R., Vitaliano, P. P., Raskind, M. A., Eisdorfer, C., Zemcuznikov, N., &
Gerber, C. J. (1982). Changes in the sleep and waking EEGs of nondemented and
demented elderly subj ects. Journal of the American Geriatric Society, 30, 86-93.

Roth T., & Roehrs T. (2003). Insomnia: epidemiology, characteristics, and consequences.
Clinical Cornerstone, 5, 5-15.

Salthouse, T. A. (1991). Theoretical perspectives on cognitive aging. Hillsdale, NJ: Erlbaum.

Salthouse, T. A., & Berish, D. E. (2005). Correlates of within-person (across-occasion)
variability in reaction time. Neuropsychology, 19, 77-87.










Salthouse, T. A., (1996). The processing-speed theory of adult age differences in cognition.
Psychological Review, 103, 403-428.

Salthouse, T. A., Nesselroade, J. R., & Berish, D. E. (2006). Short-term variability in cognitive
performance and the calibration of longitudinal change. Journal ofGerontology, 61, 144-
151.

Schneider, C., Fulda, S., & Schulz, H. (2004). Daytime variation in performance and
tiredness/sleepiness ratings in patients with insomnia, narcolepsy, sleep apnea and normal
controls. Journal of Sleep Research, 13, 373-383.

Schwartz S., Anderson, W. M., Cole, S. R., Comoni-Huntley, J., Hays, J. C., & Blazer, D.
(1999). Insomnia and heart disease: a review of epidemiologic studies. Journal of
Psychosomatic Research, 47, 313-333.

Siegler, R. S. (1994). Cognitive variability: A key to understanding cognitive development.
Current Directions in Psychological Science, 3, 1-5.

Singer, J. D., & Willett, J. B. (2003). Applied longitudinal data analysis : modeling change and
event occurrence. Oxford ; New York: Oxford University Press.

Singer, J. D., Davidson, S. M., Graham, S., & Davidson, H. S. (1998). Physician retention in
community and migrant health centers: who stays and for how long? Medical Care, 36,
1198-1213.

Singer, J. D., Fuller, B., Keiley, M. K., & Wolf, A. (1998). Early child-care selection: variation
by geographic location, maternal characteristics, and family structure. Developmental
Psychology, 34, 1129-1144.

Sliwinski, M. J., & Buschke, H. (2004). Modeling intrindividual cognitive change in aging
adults: Results from the Einstein Aging Studies. Aging Neuropsychology and Cognition,
11, 196-211.

Sliwinski, M. J., Hofer, S. M., & Hall, C. (2003). Correlated and coupled cognitive change in
older adults with and without preclinical dementia. Psychology and Aging, 18, 672-683.

Sliwinski, M. J., Smyth, J. M., Hofer, S. M., & Stawski, R. S., (2006). Intraindividual coupling
of daily stress and cognition. Psychology and Aging, 21, 545-557.

Smith, A. (1982). SynabolDigit M~oda~lities Test -Revised: Manual. Los Angeles : Westemn
Psychological Services.

Smith, A. (1991) SynabolDigit M~oda~lities Test. Los Angeles : Western Psychological Services.

Stoller, M. K. (1994). Economic effect of insomnia. Clinical Therapist, 16, 873-897.

Tabachnick, B. G., & Fidell, L. S. (2007). UsingM2dutivariate Statistics 5t Edition. Boston, MA:
Pearson Education, Inc.










Thurstone, T. G. (1962). Primary mental ability for Grades 9-12 (Rev. ed.). Chicago, 1L: Science
Research Associates.

Totterdell, P., Reynolds, S., Parkinson, B., & Briner, R. B. (1994). Associations of sleep with
everyday mood, minor symptoms and social interaction experience. Sleep, 17, 466-475.

Tworoger, S. S, Lee, S., Schernhammer, E. S., & Grodstein, F. (2006). The association of self-
reported sleep duration, difficulty sleeping, and snoring with cognitive function in older
women. Alzheimer Disease and Associated Disorders, 20, 41-48.

U. S. Census Bureau. (2004) Estimates of Average Annual Rates of the Components of
Population Change for the United States and States: April 1, 2000 to July 1, 2004 (NST-
EST2004-06)

Vallieres, A., Ivers, H., Bastien, C. H., Beaulieu-Bonneau, S. and Morin, C. M. (2005).
Variability and predictability in sleep patterns of chronic insomniacs. Journal of Sleep
Research, 14, 447-453.

Van Dongen, H. P. A., Maislin, G., & Dinges, D. F. (2004). Dealing with inter-individual
differences in the temporal dynamics of fatigue and performance: Importance and
techniques. Aviation, Space, and Environmental M~edicine, 75, 147- 154.

Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F., (2003). The cumulative
cost of additional wakefulness: does-response effects on neurobehavioral functions and
sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 26, 117-
126.

Vignola, A., Lamoureax, C., Bastien, C. H., & Morin, C. M. (2000). Effects of chronic insomnia
and use of benzodiazepines on daytime performance in older adults. Journal of
Gerontology, 55, 54-62.

Vorona, R. D., Winn, M. P., Babino, T. W., Eng, B. P., Fledman, H. R., & Ware, J. C. (2005).
Overweight and obese patients in a primary care population report less sleep than patients
with a normal body mass index. Archives oflnternal2\~edicine, 165, 25-30.

Wagner, U., Gais, S., & Born, J. (2001). Emotional memory formation is enhanced across sleep
intervals with high amounts of rapid eye movement sleep. Learning and2~emory, 8, 1 12-
119.

Wagner, u., Gais, S., Haider, H., Verleger, R., & Born, J. (2004). Sleep inspires insight. Nature,
427, 352-355.

Walker, M P., & Stickgold, R. (2004). Sleep-dependent learning and memory consolidation.
Neuron, 44, 121-133.

Walsh, J. K., & Engelhardt, C. L. (1999). Prevalence and health consequences of insomnia.
Sleep, 22, S386-S393.










Webb, W. B. (1982). Sleep in older persons: Sleep structures in 50- to 60-year-old men and
women. Journal of Gerontology, 52, 368-371.

Webb, W. B. (1985). A further analysis of age and sleep deprivation effects. Psychophysiology,
22, 156-161.

Webb, W. B., & Levy, C. M. (1982). Age, sleep deprivation, and performance.
Psychophysiology, 19, 272-276.

Wilkinson, R. T. (1992). The measurement of sleepiness. In Broughton Ogilvie (Eds.), Sleep,
arousal, and performance. Birkhauser: Sage Publications

Yesavage, J. A. (1983). Development and validation of a Geriatric Depression Screening Scale:
A preliminary report. Journal of Psychiatric Research, 1 7, 37-49.

Yoo, S. S., Hu, P. T., Gujar, N., Jolesz, F. A., & Walker, M. P. (2007). A deficit in the ability to
form new human memories without sleep. Nature Neuroscience, February 2007, 1-8.

Zepelin, H., McDonald, C. S., & Zammit, G. K. (1984). Effects of age on auditory awakening
thresholds. Journal ofGerontology, 39, 294-300.









BIOGRAPHICAL SKETCH

My scholarly interests in the aging process began when I was an undergraduate at the

University of Nevada, Las Vegas. Through witnessing the aging of my grandmother and a

seminar on cognitive aging a spark was lit in me that has never gone out. Over the course of my

junior and senior year I was awarded grants from the National Institute on Health (NIH) and the

National Science Foundation (NSF) to conduct independent research. After graduating magna

cum laude and earning a Bachelor of Arts degree in psychology I quickly applied to graduate

school. I was accepted into the Department of Clinical and Health Psychology doctoral program

at the University of Florida and awarded a competitive spot on a National Institute on Aging

Training Grant (T32). Last year I was honored to be the recipient of the American Psychological

Association's Division 20 (Adult Development and Aging) Research and Retirement Award for

most outstanding proposed master's thesis for the precursor to this document and was recently

awarded a Most Outstanding Poster award by the college of Public Health and Health

Professionals for presentation of sections of this thesis. My research interests are many, but

always share the common theme of aging. I am fascinated by the cognitive aging process, sleep's

relation to cognitive performance, and short-term variability.