Abstract
Objective
Sleep disturbances are common in older adults. Little is known about the sleep of cognitively-intact older adults and its relationship to subsequent cognitive impairment. The objective of this study was to examine the association between objective sleep-wake measures and risk of incident cognitive impairment.
Design
Prospective cohort study
Setting
Four U.S. sites
Participants
1245 women (mean age 82.6 years) without dementia participating in the Study of Osteoporotic Fractures who completed actigraphy at baseline visit and comprehensive cognitive assessment at follow-up.
Measurements
Examine the association between sleep-wake patterns measured by actigraphy and risk of incident mild cognitive impairment (MCI) and dementia.
Results
A total of 473 (38%) women developed cognitive impairment during an average (SD) follow-up of 4.9 (0.6) years; 290 (23.3%) developed MCI and 183 (14.7%) developed dementia. After controlling for multiple potential confounders, women in the lowest quartile of average sleep efficiency (<74%) had a 1.5-fold higher odds of developing MCI or dementia compared to women in the highest quartile of sleep efficiency (>86% )(OR Q1 vs. Q4 1.53, 95% CI 1.07, 2.19; Wald χ2 [1, N=1223]=5.34 for p for trend=0.03). Longer average sleep latency, but not total sleep time, was also associated with higher odds of developing cognitive impairment. Greater variability in both sleep efficiency and total sleep time was associated with an increased odds of developing MCI or dementia.
Conclusions
Lower average sleep efficiency, longer average sleep latency, and greater variability in sleep efficiency and total sleep time are associated with increased odds of developing cognitive impairment. Further research is needed to explore the mechanisms underlying these associations.
Keywords: cognitive aging, sleep disorders, older women
INTRODUCTION
Cognitive impairment and dementia represent a significant burden in the aging population; approximately 5 million adults aged 65 years and older in the United States had Alzheimer's disease dementia in 2010,(1) with the number of people with this disease expected to increase dramatically as the population ages. Significant sleep disruption is often observed in patients with dementia(2) and is believed to be related to the neurodegenerative process. Less is known about the sleep of cognitively-intact older adults and its relationship to subsequent cognitive decline. The existing prospective data examining the association between sleep disturbances and risk of incident cognitive impairment or dementia is largely based on self-report of sleep(3-8) and has yielded mixed findings, with variable associations between insomnia complaints, daytime sleepiness, total self-reported sleep time, and subsequent cognitive decline.
Evidence that objective measures of sleep quality are associated with subsequent cognitive impairment is limited. One prospective study of community dwelling older men without evidence of dementia found that lower sleep efficiency and greater nighttime wakefulness, but not total sleep time, as measured by actigraphy, were associated with subsequent decline in executive function as measured by the Trails B test.(9) Higher levels of sleep fragmentation as measured using a novel metric derived from actigraphy was associated with an increased risk of Alzheimer's disease over an average follow-up of 3.3 years in another cohort of community dwelling older adults without dementia.(10)
To test the hypothesis that poor sleep, assessed objectively using wrist actigraphy, is associated with increased risk of decline in cognitive function, we used data from 1245 older women without dementia participating in the Study of Osteoporotic Fractures, a long-term prospective cohort study. Participants completed wrist actigraphy to assess sleep-wake patterns at a baseline exam, as well as an expanded cognitive assessment an average of 4.9 years later. Specifically, we hypothesized that more fragmented sleep and short sleep are associated with increased risk of the development of mild cognitive impairment (MCI) or dementia.
METHODS
Participants
Women were participants in the Study of Osteoporotic Fractures (SOF), a longitudinal cohort study of community-dwelling women age 65 years or older, recruited from 4 study centers in Baltimore, MD; Minneapolis, MN; Portland, OR; and the Monongahela Valley near Pittsburgh, PA.(11) Women were excluded if they had a bilateral hip replacement or were unable to walk without assistance. The baseline SOF exams were conducted from 1986-1988, when 9704 Caucasian women were recruited; minority women were initially excluded due to their low incidence of hip fracture. However, in 1997-1998 662 African-American women were enrolled. For this analysis, we included participants who completed wrist actigraphy during SOF Visit 8 (2002-2004) and an expanded cognitive assessment during SOF Visit 9 (2006-2008), which included participants from 3 of the original 4 sites (MN, OR, and PA).
Of the 2570 women with actigraphy data at Visit 8 at the three clinical centers, 1785 women participated in Visit 9. Of these, 1284 provided sufficient data for adjudication of cognitive impairment status. The cognitive status for 12 of these women was adjudicated as indeterminate and they were excluded from the present analysis. We also excluded 27 women who were identified as possibly having cognitive impairment at Visit 8 based on use of dementia medications (n=8), self-report of a diagnosis of dementia (n=7), or a Mini-Mental State Examination (MMSE) score of <24 (n=16), with some overlap of criteria. The remaining 1245 women are included in the present analysis.
The institutional review boards on human research approved the study at each institution and all participating women provided informed consent.
Measures of sleep-wake patterns
Actigraphy data were collected with the Sleep-Watch-O (Sleep-Watch-O®, Ambulatory Monitoring, Inc., Ardsley, NY) actigraph, a small device worn on the nondominant wrist. Movement is measured by a piezoelectric linear accelerometer (sensitive to 0.003 g and above), which generates a voltage each time the actigraph is moved. These voltages are gathered continuously and summarized over 1-minute epochs. Actigraphy has been shown to provide a reliable estimate of sleep/wake activity.(12)
Participants were instructed to wear the actigraph continuously for a minimum of 3 days and nights (i.e., 72 hours). They were also asked to keep a sleep diary in which they recorded whether the actigraph recording represented their normal sleep/wake patterns as well as their time to bed, time of final arising, and any times the actigraphy was removed. Sleep diaries were used to aid in editing the actigraph data. Points were placed on the computer file to mark the intervals the participants were in bed trying to sleep and the times the device was removed based on time points reported on their diary. Actigraph removals listed on the diary were deleted from the analysis. If the data suggested the actigraph had been removed but no information was collected on the diary, these time points were deleted from the analysis. ActionW-2 software (Ambulatory Monitoring, Inc.) was used to analyze actigraphy data collected in the proportional integration mode.(13) This sleep scoring algorithm calculates a moving average, which takes into account the activity level immediately before and after the current minute to determine whether a given time point should be coded as sleep or wake. Details of the actigraphy scoring methods used in the SOF study have been published elsewhere.(14)
Sleep-wake parameters examined in this analysis were total sleep time per night (TST) (total hours slept while in bed), sleep latency (SL) (amount of time until sleep onset, defined as when the participant achieved sleep for 20 continuous minutes after getting into bed), and two measures of sleep fragmentation: sleep efficiency (SE) (percentage of time participant was sleeping while in bed) and time awake after sleep onset (WASO) (the total minutes of time scored as awake from sleep onset to the end of the sleep interval). For our primary analyses, a given sleep parameter averaged over all nights was used as the predictor. In addition, we examined the standard deviation of each sleep parameter over all nights as a predictor to evaluate the relationship between night-to-night variability of each sleep parameter and cognitive function.
Cognitive assessment and adjudication
The shortened Mini-Mental State Examination (MMSE)(15) ( a test of global cognition) and Trails B(16) (a test of executive function) were administered at all clinic visits including at baseline (i.e. Visit 8). At the follow-up visit (Visit 9, approximately 5 years later), an expanded neuropsychological test battery was administered, including Trails B, the Modified Mini-Mental State Examination (3MS)(17), the California Verbal Learning Test (CVLT) (second edition short form)(18), Digit Span (from the Wechsler Adult Intelligence Scale-Revised)(19), and category and verbal fluency tests.(20)
Cognitive impairment was determined in a 2-step process that has been described elsewhere.(21) Women were screened for cognitive impairment at Visit 9 using 1 or more of the following criteria: 1) score <88 on the 3MS; 2) score <4 on the California Verbal Learning Test delayed (10 minute) recall; 3) score ≥3.6 on the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE), a questionnaire designed to assess cognitive decline and dementia in elderly people completed by a relative or friend(22); 4) self-reported previous dementia diagnosis; or 5) nursing home or personal care home residence.
The women who screened positive had their cognitive status adjudicated by a panel of clinical experts, which included a neurologist, 2 neuropsychologists, and a geropsychologist, who were blinded to the actigraphy results. The remaining women who screened negative were considered to be cognitively normal. Information used for assessment included the Visit 9 neuropsychological battery scores, IQCODE, prior cognitive test scores, years of education, medical history, medications, Geriatric Depression Scale score (GDS), and functional status. The diagnosis of dementia was made based on the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DMS-IV) criteria.(23) Mild cognitive impairment (MCI) was diagnosed using a modified Petersen Criteria.(24,25) Participants were classified as having cognitive impairment if they had a diagnosis of MCI or dementia.
Other Measurements
Participants completed a questionnaire and were asked about self-reported health, education, smoking status, alcohol use, walking for exercise, and living situation. A medical history was obtained, including a history of stroke, diabetes, Parkinson's disease, non-skin cancer, chronic obstructive pulmonary disease, or coronary heart disease (angina/myocardial infarction), congestive heart failure, and hypertension. Functional status was assessed by collecting information on 6 instrumental activities of daily living (IADLs), which included walking 2 to 3 blocks on level ground, climbing up to 10 steps, walking down 10 steps, preparing meals, doing heavy housework, and shopping for groceries or clothing.(26,27)
The 15-item Geriatric Depression Scale (GDS) was used to assess depressive symptoms.(28) Body weight and height were measured, and body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters. All measurements were collected at visit 8, the baseline examination for this analysis.
Participants were asked to bring in all medications used within the preceding 30 days. All prescription and nonprescription medications were entered into an electronic database and each medication was matched to its ingredient(s) based on the Iowa Drug Information Service Drug Vocabulary (College of Pharmacy, University of Iowa, Iowa City, IA).(29) Drug categories included antidepressants, benzodiazepines, and prescription sleep medications (non-benzodiazepine, non-barbiturate sedative hypnotics).
Statistical analysis
Characteristics known to be related to sleep or cognitive function were summarized by quartile of averaged sleep efficiency across all nights using means and standard deviations for continuous data and counts and percentages for categorical data. We compared characteristics across quartiles of sleep efficiency, using analysis of variance for normally distributed continuous variables; a Kruskall-Wallis test for skewed data; and a chi-square test for categorical data. Similar comparisons were made for the other predictors examined (data not shown).
We then used logistic regression to examine the association between the objective sleep measures and risk of developing mild cognitive impairment or dementia, with a separate model for each predictor. The odds ratio (OR) and 95% confidence interval (CI) for each outcome was calculated across quartile of each sleep predictor. For averaged total sleep time, the referent group was the middle two quartiles due to the U-shaped relationship between total sleep time and cognitive function previously reported.(7,30,31) For averaged sleep efficiency, the referent group was the highest quartile; for averaged sleep latency, the referent group was the quartile with the lowest sleep latency; for averaged WASO, the referent group was the quartile with the lowest WASO. Tests for a linear trend across quartiles were performed by including each predictor (ordinal variable, four levels) as an independent variable in models. Minimally adjusted models controlled for age, race, clinic, and education. Models were further adjusted by covariates that were associated with one or more of the sleep predictors in univariate analyses or the outcome of MCI or dementia in age-adjusted models at p <0.10 (full multivariate model: age, race, clinic, education, body mass index, depressive symptoms, comorbidities, IADL impairments, smoking, alcohol, exercise, living alone, self-reported health status, use of antidepressants, benzodiazepines, and prescription sleep medication use). We also explored the association between the variability of the actigraphic measures over the days of monitoring and risk of incident MCI or dementia by examining the standard deviation of each sleep measure across all days of recording in the models. These variables were also examined as quartiles, with the lowest quartile used as the reference category.
RESULTS
Baseline characteristics
Among the 1245 women (mean age 82.6 years, SD 3.3) in the analytical sample, actigraphic data were collected for an average (SD) of 3.6 (0.7) 24-hour periods. Median (interquartile range) of average total sleep time was 6.91 (6.11-7.51) hours; median (interquartile range) averaged sleep efficiency was 81.8% (74.4-86.3%); median (interquartile range) averaged WASO was 58.3 (39.3-85.8) minutes; and median (interquartile range) averaged sleep latency was 28.8(17.3-47.5) minutes. Characteristics of participants by quartile of averaged sleep efficiency are shown in Table 1. Compared to those in higher quartiles, those with lower sleep efficiency (SE quartile 1) were on average younger and had higher BMI, more comorbidities, higher number of depressive symptoms, and more IADLS impairments. They were more likely to be African-American and to report worse health status. Characteristics of participants by quartile of total sleep time, sleep latency, and WASO appear in Supplementary Tables 1-3.
Table 1.
Baseline Characteristics of Participants for Overall Cohort by Averaged Sleep Efficiency
| Quartilea of Averaged Sleep Efficiency, % |
|||||||
|---|---|---|---|---|---|---|---|
| Characteristic | Overall (N=1245) | Q1 (N=311) | Q2 (N=311) | Q3 (N=311) | Q4 (N=312) | P-valueb | χ2 [d.f, N] or F-value [d.f] |
| Age, years, mean (SD) | 82.63 (3.32) | 82.2 (3.70) | 82.63 (3.36) | 82.91 (3.27) | 82.77 (2.85) | 0.04 | F = 2.73 [3, 1241] |
| Caucasian, n (%) | 1109 (89.08) | 243 (78.14) | 276 (88.75) | 292 (93.89) | 298 (95.51) | <0.001 | χ2 = 58.99 [3, 1245] |
| Self-reported health status, n (%) | <0.001 | χ2 = 23.76 [6, 1245] | |||||
| Poor/very poor | 16 (1.29) | 7 (2.25) | 6 (1.93) | 1 (0.32) | 2 (0.64) | ||
| Fair | 225 (18.07) | 72 (23.15) | 67 (21.54) | 40 (12.86) | 46 (14.74) | ||
| Good/excellent | 1004 (80.64) | 232 (74.60) | 238 (76.53) | 270 (86.82) | 264 (84.62) | ||
| Education, years, mean (SD) | 12.88 (2.57) | 12.73 (2.54) | 12.9 (2.54) | 12.86 (2.66) | 13.03 (2.53) | 0.54 | F = 0.71 [3, 1241] |
| Lives alone, n (%) | 734 (58.96) | 181 (58.20) | 184 (59.16) | 180 (57.88) | 189 (60.58) | 0.90 | χ2 = 0.57 [3, 1245] |
| Average drinks per week, mean (SD) | 1.12 (2.87) | 1.29 (3.67) | 1.02 (2.48) | 1.03 (2.35) | 1.16 (2.79) | 0.70 | χ2 = 1.42 [3, 1245] |
| Never smoked, n (%) | 824 (66.18) | 193 (62.06) | 200 (64.31) | 207 (66.56) | 224 (71.79) | 0.21 | χ2 = 8.46 [6, 1245] |
| Current antidepressant use, n (%) | 131 (10.52) | 42 (13.50) | 29 (9.32) | 28 (9.00) | 32 (10.26) | 0.24 | χ2 = 4.20 [3, 1245] |
| Current benzodiazepine use, n (%) | 84 (6.75) | 27 (8.68) | 19 (6.11) | 20 (6.43) | 18 (5.77) | 0.46 | χ2 = 2.57 [3, 1245] |
| Current prescription sleep medication use, n (%) | 14 (1.12) | 7 (2.25) | 2 (0.64) | 3 (0.96) | 2 (0.64) | 0.18 | χ2 = 4.92 [3, 1245] |
| # of medical conditionsc (0-8), mean (SD) | 1.39 (1.11) | 1.59 (1.16) | 1.41 (1.08) | 1.31 (1.12) | 1.25 (1.04) | <0.001 | χ2 = 16.78 [3, 1244] |
| GDS score (0-15), mean (SD) | 1.9 (2.21) | 2.34 (2.45) | 1.93 (2.29) | 1.62 (1.92) | 1.73 (2.10) | <0.001 | χ2 = 21.21 [3, 1243] |
| IADL impairments (0 - 6), mean (SD) | 1.12 (1.57) | 1.51 (1.76) | 1.11 (1.52) | 1.11 (1.59) | 0.76 (1.27) | <0.001 | χ2 = 32.33 [3, 1245] |
| BMI, kg/m2, mean (SD) | 27.5 (4.82) | 28.92 (5.26) | 27.67 (5.01) | 26.89 (4.47) | 26.53 (4.14) | <0.001 | F = 15.40 [3, 1235] |
| Walks for exercise, n (%) | 516 (41.88) | 119 (38.89) | 129 (41.75) | 142 (45.95) | 126 (40.91) | 0.34 | χ2 = 3.35 [3, 1232] |
| MMSE (0-30), mean (SD) | 28.49 (1.36) | 28.24 (1.43) | 28.48 (1.35) | 28.62 (1.36) | 28.61 (1.27) | 0.001 | χ2 = 15.97 [3, 1245] |
| Visit 9 3MS (0-100), mean (SD) | 88.10 (9.67) | 86.58 (9.83) | 88.46 (9.82) | 87.81 (10.75) | 89.56 (7.86) | <0.001 | χ2 = 19.81 [3, 1244] |
| Visit 9 cognitive status, n (%) | 0.03 | χ2 = 14.04 [6, 1245] | |||||
| Normal | 772 (62.01) | 177 (56.91) | 193 (62.06) | 190 (61.09) | 212 (67.95) | ||
| MCI | 290 (23.29) | 83 (26.69) | 63 (20.26) | 74 (23.79) | 70 (22.44) | ||
| Dementia | 183 (14.70) | 51 (16.40) | 55 (17.68) | 47 (15.11) | 30 (9.62) | ||
Abbreviations: GDS, Geriatric Depression Scale; IADL, instrumental activities of daily living; BMI, body mass index; MMSE, Mini-Mental State Examination; 3MS, Modified Mini-Mental State Examination; MCI, Mild Cognitive Impairment
Quartile cutpoints: 74.39, 81.84, 86.26
p-values for continuous variables are from ANOVA (F-statistic) for normally distributed variables, a Kruskall-Wallis test (χ2) for skewed data; p-values for categorical are data from a chi-square test for homogeneity (χ2).
Select medical conditions include history of cancer, stroke, diabetes, Parkinson's disease, chronic obstructive pulmonary disease, congestive heart failure, hypertension and coronary heart disease (angina/myocardial infarction)
911 women who did not have evidence of cognitive impairment at Visit 8 and were alive at Visit 9 did not complete the Visit 9 cognitive assessment and therefore were not included in this analysis. Characteristics of these women compared to the women in our analytic cohort appear in Supplementary Table 4. Women not included in this analysis were older on average by about one year, had a slightly lower MMSE score on average, and did have lower average sleep efficiency, higher average WASO, and higher average variability in total sleep time and sleep efficiency.
Averaged sleep measures and MCI/dementia
A total of 473 (38%) women developed cognitive impairment during an average (SD) follow-up of 4.9 (0.6) years; 290 (23.3%) developed MCI and 183 (14.7%) developed dementia.
Lower averaged sleep efficiency was associated with greater odds of developing MCI or dementia (Table 2). In a minimally adjusted model (covariates of age, race, clinic, and education), women in quartile (Q) 1 (<74.39%) had a 1.55-fold higher risk of developing MCI or dementia compared with women in Q4 (referent group, ≥86.261). After adjustment for multiple potential confounders in the full model, (i.e., age, race, clinic, education, BMI, number of depressive symptoms, comorbidities, number of IADL impairments, smoking status, alcohol use, exercise, living alone, self-reported health status, antidepressant use, benzodiazepine use, and prescription sleep medication use), results were similar: women in Q1 compared to Q4 had a 1.53 higher risk of developing cognitive impairment. Longer sleep latency was also associated with an increased odds of developing MCI or dementia. In the minimally adjusted model, women in quartile 4 (>47.5 min) had a 1.4-fold higher risk of developing MCI or dementia compared with women in Q1 (referent group, <17.25 min). In the full multivariate model, results were similar similar (Wald χ2 [1, N=1223]=4.07 for p for trend=0.04). Total sleep time was not associated with the odds of developing cognitive impairment in either the minimally adjusted model or full multivariate model, nor was the number of minutes of wakefulness after sleep onset (Table 2).
Table 2.
Averaged Objective Sleep Measures and Odds of Dementia or MCI
| Outcome = MCI or Dementia vs. Normal |
||||
|---|---|---|---|---|
| Minimally Adjusteda | Multivariate Adjustedb | |||
| Predictor (average over all nights) | OR (95% CI) | Wald χ2 [1, N=1245], p-value | OR (95% CI) | Wald χ2 [1, N=1223], p-value |
| Total Sleep Time, hours | ||||
| Q1 (<6 hrs, 7 min) | 0.92 (0.69-1.24) | 0.30, p = 0.59 | 0.85 (0.63-1.16) | 1.02, p = 0.31 |
| Q2-3 (6 hrs to <7.5 hrs) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q4 (≥7.5 hrs) | 1.00 (0.75-1.33) | 0.00, p = 0.98 | 0.93 (0.69-1.25) | 0.26, p = 0.61 |
| Sleep Efficiency, % | ||||
| Q1 (<74.39) | 1.55 (1.10-2.18) | 6.28, p = 0.01 | 1.53 (1.07-2.19) | 5.34, p = 0.02 |
| Q2 (74.39 to <81.84) | 1.31 (0.93-1.83) | 2.37, p = 0.12 | 1.33 (0.94-1.89) | 2.53, p = 0.11 |
| Q3 (81.84 to <86.26) | 1.34 (0.96-1.88) | 2.91, p = 0.09 | 1.38 (0.97-1.95) | 3.25, p = 0.07 |
| Q4 (≥86.26) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| p-trend | 5.40, p = 0.02 | 4.54, p = 0.03 | ||
| Wake After Sleep Onset, minutes | ||||
| Q1 (<39.25) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q2 (39.25 to <58.33) | 1.21 (0.87-1.70) | 1.30, p = 0.26 | 1.27 (0.90-1.80) | 1.87, p = 0.17 |
| Q3 (58.33 to <85.75) | 1.16 (0.83-1.62) | 0.75, p = 0.39 | 1.14 (0.80-1.61) | 0.51, p = 0.47 |
| Q4 (≥85.75) | 1.21 (0.86-1.69) | 1.18, p = 0.28 | 1.16 (0.81-1.66) | 0.66, p = 0.42 |
| p-trend | 0.89, p = 0.35 | 0.33, p = 0.57 | ||
| Sleep Latency, minutes | ||||
| Q1 (<17.25) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q2 (17.25 to <28.75) | 1.08 (0.77-1.52) | 0.21, p = 0.65 | 1.12 (0.79-1.59) | 0.41, p = 0.52 |
| Q3 (28.75 to <47.50) | 1.29 (0.92-1.80) | 2.16, p = 0.14 | 1.37 (0.97-1.93) | 3.19, p = 0.07 |
| Q4 (≥47.50) | 1.42 (1.01-1.99) | 4.14, p = 0.04 | 1.37 (0.96-1.94) | 3.04, p = 0.08 |
| p-trend | 5.06, p = 0.03 | 4.07, p = 0.04 | ||
Adjusted for age, race, clinic and education
Adjusted for age, race, clinic, education, body mass index, number of depressive symptoms, comorbidities, number of IADL impairments, smoking status, alcohol use, exercise, living alone, self-reported health status, antidepressant use, benzodiazepine use, and prescription sleep medication use.
A separate model was run for each predictor.
Variability in sleep measures and MCI/dementia
Greater variability in sleep efficiency was associated with risk of developing MCI and dementia (Table 3). In both a minimally adjusted model and the full multivariate model, women in Q4 (most variability in sleep efficiency over the all nights) had a 1.9-fold higher risk of developing MCI or dementia compared to those in the lowest quartile (full multivariate model). Variability in total sleep time was also associated with an increased risk of developing MCI or dementia: women in Q4 of total sleep time variability had a 1.4-fold higher risk compared to those in the lowest quartile of variability for the full multivariate model. Variability in time spent awake after sleep onset (WASO) and sleep latency were not related to risk of developing MCI and dementia.
Table 3.
Variability in Objective Sleep Measures and Odds of MCI or Dementia
| Outcome = MCI or Dementia vs. Normal |
||||
|---|---|---|---|---|
| Minimally Adjusteda | Multivariate Adjustedb | |||
| Predictor (SD over all nights) | OR (95% CI) | Wald χ2 [1, N=1242], p-value | OR (95% CI) | Wald χ2 [1, N=1220], p-value |
| Total Sleep Time, minutes | ||||
| Q1 (<28.58) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q2 (28.58 to <42.16) | 0.81 (0.58-1.14) | 1.49, p = 0.22 | 0.82 (0.58-1.17) | 1.21, p = 0.27 |
| Q3 (42.16 to <61.04) | 0.96 (0.68-1.34) | 0.07, p = 0.79 | 0.98 (0.69-1.38) | 0.02, p = 0.89 |
| Q4 (≥61.04) | 1.36 (0.97-1.91) | 3.26, p = 0.07 | 1.40 (0.98-1.98) | 3.47, p = 0.06 |
| p-trend | 4.02, p = 0.045 | 4.05, p = 0.04 | ||
| Sleep Efficiency, % | ||||
| Q1 (<2.99) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q2 (2.99 to <4.93) | 1.83 (1.30-2.57) | 12.00, p = 0.0005 | 1.89 (1.33-2.69) | 12.50, p = 0.0004 |
| Q3 (4.93 to <7.92) | 1.43 (1.01-2.02) | 4.11, p = 0.04 | 1.48 (1.04-2.12) | 4.60, p = 0.03 |
| Q4 (≥7.92) | 1.87 (1.32-2.64) | 12.65, p = 0.0004 | 1.92 (1.34-2.75) | 12.74, p = 0.0004 |
| p-trend | 8.40, p = 0.004 | 8.66, p = 0.003 | ||
| Wake After Sleep Onset, minutes | ||||
| Q1 (<11.64) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q2 (11.64 to <20.33) | 1.31 (0.94-1.83) | 2.52, p = 0.11 | 1.29 (0.91-1.82) | 2.10, p = 0.15 |
| Q3 (20.33 to <37.44) | 1.21 (0.86-1.69) | 1.17, p = 0.28 | 1.22 (0.86-1.73) | 1.26, p = 0.26 |
| Q4 (≥37.44) | 1.24 (0.88-1.75) | 1.54, p = 0.21 | 1.25 (0.87-1.78) | 1.48, p = 0.22 |
| p-trend | 1.04, p = 0.31 | 1.11, p = 0.29 | ||
| Sleep Latency, minutes | ||||
| Q1 (<10.10) | 1.00 (reference) | - - | 1.00 (reference) | - - |
| Q2 (10.10 to <18.15) | 1.00 (0.71-1.40) | 0.00, p = 1.00 | 0.98 (0.69-1.38) | 0.02, p = 0.90 |
| Q3 (18.15 to <33.89) | 0.94 (0.67-1.32) | 0.11, p = 0.74 | 0.92 (0.65-1.31) | 0.20, p = 0.66 |
| Q4 (≥33.89) | 1.36 (0.98-1.91) | 3.30, p = 0.07 | 1.37 (0.97-1.94) | 3.20, p = 0.07 |
| p-trend | 2.66, p = 0.10 | 2.60, p = 0.11 | ||
Adjusted for age, race, clinic and education
Adjusted for age, race, clinic, education, body mass index, number of depressive symptoms, comorbidities, number of IADL impairments, smoking status, alcohol use, exercise, living alone, self-reported health status, antidepressant use, benzodiazepine use, and prescription sleep medication use.
A separate model was run for each predictor.
CONCLUSIONS
We found that lower averaged sleep efficiency and longer averaged sleep latency, but not total sleep time or nighttime wakefulness, were associated with a higher odds of developing cognitive impairment. These findings persisted after controlling for multiple potential confounders. Women with the lowest sleep efficiency had a 1.5 greater odds of developing MCI or dementia over the 4.9 years of follow-up, while women with the longer sleep latencies had a 1.4 greater odds of cognitive decline. In addition, we found that variability in sleep efficiency and total sleep time were also associated with increased odds of developing cognitive impairment.
Older adults with dementia have been well documented to experience disturbed sleep,(32-34) thought to be due to changes associated with the neurodegenerative process, as well as reflective of comorbidities.(35,36) Sleep is disturbed early in the neurodegenerative process and sleep disturbances are observed in the presence of MCI.(2,37) However, it is more uncertain if sleep disturbances increase risk for future development of cognitive impairment.
Potential biologic links between sleep disturbances and cognitive decline have been identified. The accumulation of amyloid-beta in the brain extracellular space is an important event in the pathogenesis of AD. Both the sleep-wake cycle and orexin, a neurotransmitter that regulates wakefulness, have been shown to play a role in regulating amyloid-beta dynamics.(38) The effects of sleep deprivation and orexin modulation have been examined in wild-type mice and human amyloid precursor protein transgenic mice, which express a mutated form of hAPP.(38) The quantity of amyloid-beta measured in brain interstitial fluid was shown to correlate with the amount of wakefulness and with orexin dynamics.(38) Chronic sleep restriction increased amyloid-B plaque formation in the amyloid precursor protein transgenic mice.(38) Other recent work suggests that sleep affects the ability of the brain to remove potentially toxic biomolecules that accumulate with normal neuronal function, including beta-amyloid.(39) Clearance of these toxic waste products is as much as two-fold faster in sleep than in waking hours.(39) In addition, in aged rats, chronic sleep restriction leads to alterations in Ca++ signaling and abnormalities of synaptic plasticity in the hippocampus, an area of the brain that plays an important role in memory and where early abnormalities are seen in Alzheimer's disease.(40-42) Recent studies in humans have linked lower sleep efficiency and greater WASO measured by actigraphy as well as poorer sleep quality and shorter sleep duration measured by self-report to greater amyloid deposition.(43,44)
In humans, cross-sectional studies have reported associations between sleep disruption and cognitive impairments in community-dwelling older adults. Of those that have examined objective measures of sleep and cognitive function, sleep fragmentation, as measured by a variety of parameters, has been associated with poorer cognition.(45-47) However, due to the cross-sectional nature of this work, conclusions regarding the direction of these associations cannot be made. Much of the limited prospective data examining the association between sleep disturbances and risk of incident cognitive impairment have focused on self-reported total sleep time (TST) and sleep quality.(3-8) One prospective study of community dwelling older adults without dementia that did objectively measure sleep utilizing actigraphy found that higher levels of sleep fragmentation were associated with an increased risk of Alzheimer's disease(10); however, that study did not measure other sleep parameters, such as sleep efficiency, WASO, or total sleep time. Another prospective study(9) found that lower sleep efficiency and greater nighttime wakefulness as measured by actigraphy was associated with an increased risk of subsequent decline on a measure of executive function; however, this work did not measure outcomes of mild cognitive impairment or dementia, as we did in the present study.
Although we did observe an association between sleep efficiency and incident cognitive impairment, we did not observe an association between total sleep time and incident cognitive impairment. Animal work has suggested that decreased total sleep time may potentially play a role in the accumulation of beta-amyloid in the brain(38,39) and Spira et al(44) found that shorter self-reported total sleep time was associated with greater beta-amyloid burden cross-sectionally. Our prospective data based on actigraphic measures suggests that reduced sleep efficiency, rather than total sleep time, is more strongly associated with incident cognitive impairment.
We found that variability in sleep efficiency and total sleep time were also associated with increased odds of developing MCI or dementia. Greater variability in sleep efficiency and WASO has been correlated with poorer performance on a new associative learning task(48) in older adults. Little work to date has examined variability in these sleep parameters. However, increased daily variation in these parameters may be tied to disruptions in the alignment of circadian rhythms, which have been associated with a range of health outcomes.(49) Future studies should explore whether this variability is an early symptom of cognitive changes and whether such variability reflects weakening of circadian rhythms.
Lower sleep efficiency and increased variability of sleep parameters in adults without significant cognitive impairment may be associated with the development of later cognitive impairment due to early neurodegenerative changes not yet manifesting as cognitive impairment. Alternatively, these alterations in sleep may have a causal relationship with the later development of cognitive impairment. Further research will be important to better identify the mechanisms underlying this association.
This research has several strengths. We examined a large cohort of community-dwelling women who were not selected on the basis of sleep or cognitive complaints. We examined objective measures of sleep using actigraphy, adjusted for multiple important confounders, and utilized a rigorous adjudication process for the outcomes of MCI and dementia. We excluded participants with probable dementia or cognitive impairment at Visit 8, the baseline visit for this analysis, and so were able to evaluate for risk of incident MCI and dementia. There are, however, several limitations. The cognitive assessment at Visit 8 was less robust than at follow-up and, as a result, we cannot exclude the possibility that some women included in the analysis had MCI or dementia at Visit 8. If women who had cognitive impairment at Visit 8 were included in this analysis, it is possible that they already were experiencing changes in their sleep-wake cycle and therefore could have biased our results. In addition, to be included in our analysis, women had to have completed actigraphy at Visit 8 and a cognitive assessment at Visit 9; women who did not complete Visit 9 may have been more likely to have had cognitive impairment at follow-up. On average, these women not included in the analysis had lower sleep efficiency, higher WASO, and more variability in their sleep parameters than those women included in the analysis, potentially attenuating our observed association between actigraphic measurements and incident cognitive impairment. Due to the observational nature of the study, we cannot exclude residual confounding that may explain our observed results. Lastly, because the cohort was restricted to older, primarily Caucasian women, we cannot generalize our findings to other populations, such as men, younger women, or a more diverse population.
Further research will be important to better identify the mechanisms underlying the observed associations between these sleep parameters and the development of MCI and dementia. Identifying older adults with sleep disturbances may be useful prognostically to identify those at higher risk for cognitive impairment. Future work could examine whether interventions to improve sleep efficiency and reduce variability in sleep parameters improve cognitive outcomes in older adults.
Supplementary Material
Acknowledgments
Source of Funding
The Study of Osteoporotic Fractures (SOF) is supported by National Institutes of Health funding. The National Institute on Aging (NIA) provides support under the following grant numbers: R01 AG005407, R01 AR35582, R01 AR35583, R01 AR35584, R01 AG005394, R01 AG027574, and R01 AG027576. The funding agencies had no direct role in the conduct of the study; the collection, management, analyses and interpretation of the data; or preparation or approval of the manuscript.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest: All authors have nothing to disclose.
Supplemental Digital Content
Supplementary Tables 1-4 (15-0626 SOF actigraphy and cognitive decline_suppl tables AJGP.docx)
REFERENCES
- 1.Hebert LE, Weuve J, Scherr PA, et al. Alzheimer disease in the United States (2010-2050) estimated using the 2010 census. Neurology. 2013;80(19):1778–1783. doi: 10.1212/WNL.0b013e31828726f5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Vitiello MV, Prinz PN. Alzheimer's disease. Sleep and sleep/wake patterns. Clin Geriatr Med. 1989;5(2):289–299. [PubMed] [Google Scholar]
- 3.Cricco M, Simonsick EM, Foley DJ. The impact of insomnia on cognitive functioning in older adults. J Am Geriatr Soc. 2001;49(9):1185–1189. doi: 10.1046/j.1532-5415.2001.49235.x. [DOI] [PubMed] [Google Scholar]
- 4.Foley D, Monjan A, Masaki K, et al. Daytime sleepiness is associated with 3-year incident dementia and cognitive decline in older Japanese-American men. J Am Geriatr Soc. 2001;49(12):1628–1632. doi: 10.1046/j.1532-5415.2001.t01-1-49271.x. [DOI] [PubMed] [Google Scholar]
- 5.Jelicic M, Bosma H, Ponds RW, et al. Subjective sleep problems in later life as predictors of cognitive decline. Report from the Maastricht Ageing Study (MAAS). Int J Geriatr Psychiatry. 2002;17(1):73–77. doi: 10.1002/gps.529. [DOI] [PubMed] [Google Scholar]
- 6.Keage HA, Banks S, Yang KL, et al. What sleep characteristics predict cognitive decline in the elderly? Sleep Med. 2012;13(7):886–892. doi: 10.1016/j.sleep.2012.02.003. [DOI] [PubMed] [Google Scholar]
- 7.Potvin O, Lorrain D, Forget H, et al. Sleep quality and 1-year incident cognitive impairment in community-dwelling older adults. Sleep. 2012;35(4):491–499. doi: 10.5665/sleep.1732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Tworoger SS, Lee S, Schernhammer ES, et al. The association of self-reported sleep duration, difficulty sleeping, and snoring with cognitive function in older women. Alzheimer Dis Assoc Disord. 2006;20(1):41–48. doi: 10.1097/01.wad.0000201850.52707.80. [DOI] [PubMed] [Google Scholar]
- 9.Blackwell T, Yaffe K, Laffan A, et al. Associations of objectively and subjectively measured sleep quality with subsequent cognitive decline in older community-dwelling men: the MrOS sleep study. Sleep. 2014;37(4):655–663. doi: 10.5665/sleep.3562. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lim AS, Kowgier M, Yu L, et al. Sleep Fragmentation and the Risk of Incident Alzheimer's Disease and Cognitive Decline in Older Persons. Sleep. 2013;36(7):1027–1032. doi: 10.5665/sleep.2802. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cummings SR, Nevitt MC, Browner WS, et al. Risk factors for hip fracture in white women. Study of Osteoporotic Fractures Research Group. N Engl J Med. 1995;332(12):767–773. doi: 10.1056/NEJM199503233321202. [DOI] [PubMed] [Google Scholar]
- 12.Ancoli-Israel S, Cole R, Alessi C, et al. The role of actigraphy in the study of sleep and circadian rhythms. Sleep. 2003;26(3):342–392. doi: 10.1093/sleep/26.3.342. [DOI] [PubMed] [Google Scholar]
- 13.Action-W User's Guide, Version 2.0. Ambulatory Monitoring, Inc.; Ardsley, NY: 1999. [Google Scholar]
- 14.Blackwell T, Ancoli-Israel S, Gehrman PR, et al. Actigraphy scoring reliability in the study of osteoporotic fractures. Sleep. 2005;28(12):1599–1605. doi: 10.1093/sleep/28.12.1599. [DOI] [PubMed] [Google Scholar]
- 15.Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–198. doi: 10.1016/0022-3956(75)90026-6. [DOI] [PubMed] [Google Scholar]
- 16.Reitan RM, Wolfson D. The Halstead-Reitan Neuropsychological Battery: Theory and Clinical Interpretation. Neuropsychology Press; Tuscon, AZ: 1985. [Google Scholar]
- 17.Teng EL, Chui HC. The Modified Mini-Mental State (3MS) examination. J Clin Psychiatry. 1987;48(8):314–318. [PubMed] [Google Scholar]
- 18.Delis DC, Kramer JH, Kaplan E, et al. California Verbal Learning Test - Second Edition (CVLT-II) Psychological Corporation; San Antonio, TX: 2000. [Google Scholar]
- 19.Wechsler D. Wechsler Adult Intelligence Scale - Revised. Oxford University Press; New York, NY: 1991. [Google Scholar]
- 20.Spreen O, Strauss E. A Compendium of Neuropsychological Test: Administration, Norms, and Commentary. Oxford University Press; New York, NY: 1991. [Google Scholar]
- 21.Yaffe K, Middleton LE, Lui LY, et al. Mild cognitive impairment, dementia, and their subtypes in oldest old women. Arch Neurol. 2011;68(5):631–636. doi: 10.1001/archneurol.2011.82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Jorm AF, Jacomb PA. The Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE): socio-demographic correlates, reliability, validity and some norms. Psychol Med. 1989;19(4):1015–1022. doi: 10.1017/s0033291700005742. [DOI] [PubMed] [Google Scholar]
- 23.American Psychiatric Association . Diagnostic and Statistical Manual of Mental Disorders. American Psychiatric Association; Washington, D.C.: 2000. [Google Scholar]
- 24.Petersen RC, Smith GE, Waring SC, et al. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol. 1999;56(3):303–308. doi: 10.1001/archneur.56.3.303. [DOI] [PubMed] [Google Scholar]
- 25.Petersen RC, Doody R, Kurz A, et al. Current concepts in mild cognitive impairment. Arch Neurol. 2001;58(12):1985–1992. doi: 10.1001/archneur.58.12.1985. [DOI] [PubMed] [Google Scholar]
- 26.Fitti JE, Kovar MG. The Supplement on Aging to the 1984 National Health Interview Survey. Vital Health Stat. 1987;1(21):1–115. [PubMed] [Google Scholar]
- 27.Pincus T, Summey JA, Soraci SA, Jr., et al. Assessment of patient satisfaction in activities of daily living using a modified Stanford Health Assessment Questionnaire. Arthritis Rheum. 1983;26(11):1346–1353. doi: 10.1002/art.1780261107. [DOI] [PubMed] [Google Scholar]
- 28.Sheikh J, Yesavage JA. Geriatric Depression Scale (GDS): recent evidence and development of a shorter version. In: Brink TL, editor. Clinical Gerontology: a guide to assessment and intervention. The Haworth Press; New York: 1986. pp. 165–173. [Google Scholar]
- 29.Pahor M, Chrischilles EA, Guralnik JM, et al. Drug data coding and analysis in epidemiologic studies. Eur J Epidemiol. 1994;10(4):405–411. doi: 10.1007/BF01719664. [DOI] [PubMed] [Google Scholar]
- 30.Ferrie JE, Shipley MJ, Akbaraly TN, et al. Change in sleep duration and cognitive function: findings from the Whitehall II Study. Sleep. 2011;34(5):565–573. doi: 10.1093/sleep/34.5.565. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kronholm E, Sallinen M, Suutama T, et al. Self-reported sleep duration and cognitive functioning in the general population. J Sleep Res. 2009;18(4):436–446. doi: 10.1111/j.1365-2869.2009.00765.x. [DOI] [PubMed] [Google Scholar]
- 32.Bombois S, Derambure P, Pasquier F, et al. Sleep disorders in aging and dementia. J Nutr Health Aging. 2010;14(3):212–217. doi: 10.1007/s12603-010-0052-7. [DOI] [PubMed] [Google Scholar]
- 33.Moe KE, Vitiello MV, Larsen LH, et al. Symposium: Cognitive processes and sleep disturbances: Sleep/wake patterns in Alzheimer's disease: relationships with cognition and function. J Sleep Res. 1995;4(1):15–20. doi: 10.1111/j.1365-2869.1995.tb00145.x. [DOI] [PubMed] [Google Scholar]
- 34.Prinz PN, Vitaliano PP, Vitiello MV, et al. Sleep, EEG and mental function changes in senile dementia of the Alzheimer's type. Neurobiol Aging. 1982;3(4):361–370. doi: 10.1016/0197-4580(82)90024-0. [DOI] [PubMed] [Google Scholar]
- 35.Lim AS, Ellison BA, Wang JL, et al. Sleep is related to neuron numbers in the ventrolateral preoptic/intermediate nucleus in older adults with and without Alzheimer's disease. Brain. 2014;137(Pt 10):2847–2861. doi: 10.1093/brain/awu222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Montplaisir J, Petit D, Lorrain D, et al. Sleep in Alzheimer's disease: further considerations on the role of brainstem and forebrain cholinergic populations in sleep-wake mechanisms. Sleep. 1995;18(3):145–148. doi: 10.1093/sleep/18.3.145. [DOI] [PubMed] [Google Scholar]
- 37.Beaulieu-Bonneau S, Hudon C. Sleep disturbances in older adults with mild cognitive impairment. Int Psychogeriatr. 2009;21(4):654–666. doi: 10.1017/S1041610209009120. [DOI] [PubMed] [Google Scholar]
- 38.Kang JE, Lim MM, Bateman RJ, et al. Amyloid-beta dynamics are regulated by orexin and the sleep-wake cycle. Science. 2009;326(5955):1005–1007. doi: 10.1126/science.1180962. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Xie L, Kang H, Xu Q, et al. Sleep drives metabolite clearance from the adult brain. Science. 2013;342(6156):373–377. doi: 10.1126/science.1241224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.de Souza L, Smaili SS, Ureshino RP, et al. Effect of chronic sleep restriction and aging on calcium signaling and apoptosis in the hippocampus of young and aged animals. Prog Neuropsychopharmacol Biol Psychiatry. 2012;39(1):23–30. doi: 10.1016/j.pnpbp.2012.01.018. [DOI] [PubMed] [Google Scholar]
- 41.Tartar JL, Ward CP, McKenna JT, et al. Hippocampal synaptic plasticity and spatial learning are impaired in a rat model of sleep fragmentation. Eur J Neurosci. 2006;23(10):2739–2748. doi: 10.1111/j.1460-9568.2006.04808.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Tartar JL, McKenna JT, Ward CP, et al. Sleep fragmentation reduces hippocampal CA1 pyramidal cell excitability and response to adenosine. Neurosci Lett. 2010;469(1):1–5. doi: 10.1016/j.neulet.2009.11.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Ju YE, McLeland JS, Toedebusch CD, et al. Sleep quality and preclinical Alzheimer disease. JAMA Neurol. 2013;70(5):587–593. doi: 10.1001/jamaneurol.2013.2334. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Spira AP, Gamaldo AA, An Y, et al. Self-reported sleep and beta-amyloid deposition in community-dwelling older adults. JAMA Neurol. 2013;70(12):1537–1543. doi: 10.1001/jamaneurol.2013.4258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Blackwell T, Yaffe K, Ancoli-Israel S, et al. Poor sleep is associated with impaired cognitive function in older women: the study of osteoporotic fractures. J Gerontol A Biol Sci Med Sci. 2006;61(4):405–410. doi: 10.1093/gerona/61.4.405. [DOI] [PubMed] [Google Scholar]
- 46.Blackwell T, Yaffe K, Ancoli-Israel S, et al. Association of sleep characteristics and cognition in older community-dwelling men: the MrOS sleep study. Sleep. 2011;34(10):1347–1356. doi: 10.5665/SLEEP.1276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Oosterman JM, van Someren EJ, Vogels RL, et al. Fragmentation of the rest-activity rhythm correlates with age-related cognitive deficits. J Sleep Res. 2009;18(1):129–135. doi: 10.1111/j.1365-2869.2008.00704.x. [DOI] [PubMed] [Google Scholar]
- 48.Sherman SM, Schnyer DM. The contribution of sleep variability to memory performance in older adults. Sleep. 2013;36(Abstract Suppl):A80–A81. [Google Scholar]
- 49.Tranah GJ, Blackwell T, Ancoli-Israel S, et al. Circadian activity rhythms and mortality: the study of osteoporotic fractures. J Am Geriatr Soc. 2010;58(2):282–291. doi: 10.1111/j.1532-5415.2009.02674.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.