You are on page 1of 8

ORIGINAL INVESTIGATION

Actigraphy-Measured Sleep Characteristics and Risk of Falls in Older Women


Katie L. Stone, PhD; Sonia Ancoli-Israel, PhD; Terri Blackwell, MA; Kristine E. Ensrud, MD; Jane A. Cauley, DrPH; Susan Redline, MD; Teresa A. Hillier, MD; Jennifer Schneider, MPH; David Claman, MD; Steven R. Cummings, MD

Background: Prior studies have suggested that

Results: In multivariate-adjusted models, relative to those

insomnia and self-reported poor sleep are associated with increased risk of falls. However, no previous study, to our knowledge, has tested the independent associations of objectively estimated characteristics of sleep and risk of falls, accounting for the use of commonly prescribed treatments for insomnia.
Methods: Study subjects were participants in the Study of Osteoporotic Fractures. In 2978 primarily community-dwelling women 70 years and older (mean age, 84 years), sleep and daytime inactivity were estimated using wrist actigraphy data collected for a minimum of 3 consecutive 24-hour periods (mean duration, 86.3 hours). Fall frequency during the subsequent year was ascertained by a triannual questionnaire. Use of medications was obtained by examiner interview.

with normal nighttime sleep duration ( 7 to 8 hours per night), the odds of having 2 or more falls in the subsequent year was elevated for women who slept 5 hours or less per night (odds ratio, 1.52; 95% confidence interval, 1.03-2.24). This association was not explained by the use of benzodiazepines. Indexes of sleep fragmentation were also associated with an increased risk of falls. For example, women with poor sleep efficiency ( 70% of time in bed spent sleeping) had 1.36-fold increased odds of falling compared with others (odds ratio, 1.36; 95% confidence interval, 1.07-1.74).
Conclusion: Short nighttime sleep duration and in-

creased sleep fragmentation are associated with increased risk of falls in older women, independent of benzodiazepine use and other risk factors for falls. Arch Intern Med. 2008;168(16):1768-1775 falls 5 or incomplete covariate assessment.5,6 Older adults with insomnia are more likely to use hypnotic medications such as benzodiazepines, which have been associated with an increased risk of falls and fractures.7-11 However, sleep characteristics have seldom been considered as covariates in these prior studies. It is not established whether it is poor sleep or medications used to treat sleep disturbances that explain the increased risk of falls in those who are prescribed such medications. To our knowledge, the present study is the first to examine the relationship between objective estimates of sleep duration and fragmentation and subsequent risk of recurrent falls. The SOF provides a unique opportunity to study this question in a large cohort of primarily community-dwelling older women, who are well characterized for sleep exposures as well as critical covariates such as use of hypnotics and other medica-

Author Affiliations are listed at the end of this article.

risk among older adults and are a leading cause of mortality, morbidity, and premature nursing home placement.1,2 It is estimated that falls occur in approximately a third of persons older than 65 years each year.3 Insomnia and disturbed sleep are also increasingly common in older adults. A few prior studies have found that insomnia or other sleep characteristics are related to an increased risk of falls.4-6 A recently published study from our group, based on data from the multicenter Study of Osteoporotic Fractures (SOF), found that selfreported daily napping predicts the risk of incident falls and fractures.4 In addition, both daily napping and self-reported long 24-hour sleep duration were associated with an increased risk of incident fractures. All studies thus far have been limited by the use of subjective sleep data,4-6 and most have had other limitations, including retrospective ascertainment of

ALLS POSE A MAJOR HEALTH

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1768

WWW.ARCHINTERNMED.COM

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

tions, neuropsychiatric and physical function, and comorbidities.


METHODS

the same night, the intraclass correlation was highest for total sleep time (0.76) and moderately high for sleep efficiency and wake time after sleep onset (0.61 and 0.58, respectively).13

ASCERTAINMENT OF INCIDENT FALLS


Participants were contacted by postcard or telephone every 4 months from baseline to ascertain vital status, self-reported incident falls, and the occurrence of fractures. Information from designated proxy sources (eg, a family member or close friend) was used if the participant was unable to respond. After more than 16 years of follow-up since the baseline examination, these contacts remain more than 95% complete. A fall was defined as falling all the way down to the floor or ground, or falling and hitting an object like a chair or stair. All falls reported on the first 3 triannual postcards returned after the eighth examination were included in this analysis; the mean (SD) follow-up for falls was 12.1 (0.9) months. The primary outcome for this analysis was defined as the occurrence of 2 or more falls (vs fewer than 2 falls) in the year after sleep assessment.

SUBJECTS
The SOF is a prospective study originally designed to determine risk factors for osteoporotic fractures in older women. Between September 1986 and October 1988, 9704 communitydwelling white women 65 years and older were recruited from population-based listings in Baltimore, Maryland; Minneapolis, Minnesota; Portland, Oregon; and the Monongahela Valley near Pittsburgh, Pennsylvania. At baseline, women were excluded if they were unable to walk without assistance or if they had undergone a previous bilateral hip replacement. Initially, African American women were excluded because of the low incidence of osteoporotic fracture among them. However, between February 1997 and February 1998, an additional 662 African American women 65 years and older were enrolled. Further details of the SOF cohort have been published elsewhere.12 From January 2002 through April 2004, all surviving participants were invited to attend the eighth examination. Of 3676 women who participated in the eighth examination, the analysis subset comprised 2978 women who completed actigraphy and had at least 1 year of follow-up for incident falls subsequent to the examination. Of those excluded, 127 died within 1 year after the eighth examination, 568 were missing actigraphy data, and 3 had incomplete falls data. All women provided written informed consent, and the study was approved by the institutional review board at each site.

OTHER MEASUREMENTS
A comprehensive examination included measurements of height using a Harpenden stadiometer (Holtain Ltd, Croswell, Crymych, United Kingdom), and weight using a balance beam scale. Body mass index was calculated as weight in kilograms divided by height in meters squared. Tests of physical function included grip strength using a handheld isometric dynamometer ( Jamar Inc, Jackson, Mississippi), walk speed (time in seconds to walk 6 m at usual pace), and ability to complete 5 stands from a chair without use of arms or other assistance. During an interview by a trained examiner, a complete inventory of current medication use (including use of benzodiazepines, benzodiazepine receptor agonists, antipsychotics, and antidepressants) was recorded and confirmed by examination of pill bottles. Medications were categorized using a computerized medication coding dictionary.16 Depressive symptoms were assessed using the 15-item Geriatric Depression Scale; depression was defined as a Geriatric Depression Scale score of 6 or higher.17,18 Anxiety was measured by the 9-item Goldberg Anxiety Scale, with a score of 6 or higher indicating anxiety symptoms.19 The Mini-Mental State Examination20 was administered. This is a brief, global cognitive function test with concentration, language, and memory components designed to screen for cognitive impairment. The Mini-Mental State Examination scale ranges from 0 to 30, with higher numbers indicating better performance. To assess functional disability, women were asked whether they had any difficulty performing any of 6 instrumental activities of daily living (IADL), including ability to walk 2 to 3 blocks on level ground, climb up 10 steps, walk down 10 steps, prepare meals, do heavy housework, and shop for groceries or clothing.21 All participants completed a questionnaire with information on physical activity (self-reported walking for exercise) and consumption of alcohol. Medical information included a history of a physician diagnosis of cardiovascular disease (including myocardial infarction, angina, congestive heart failure, other heart disease), stroke, diabetes, Parkinson disease, chronic obstructive pulmonary disease, and cancer. An indicator variable was created to indicate prevalence of 1 or more medical conditions. Possible dementia was defined as a Mini-Mental State Examination score lower than 26, self-reported history of dementia, or use of medications commonly prescribed for dementia. Daytime sleepiness was defined as a score higher than 10 on the Epworth Sleepiness Scale.22,23 Participants were also asked
WWW.ARCHINTERNMED.COM

WRIST ACTIGRAPHY
Objective characteristics of sleep were estimated using an actigraph (SleepWatch-O; Ambulatory Monitoring Inc, Ardsley, New York). The actigraphs were worn on the nondominant wrist for a minimum of 72 hours (mean [SD] duration, 86.3 [19.3] hours). The actigraph is similar in size and weight to a standard wristwatch, and movement is detected via a piezoelectric bimorph-ceramic cantilever that generates a voltage each time the actigraph is moved. These voltages are gathered continuously and summarized over 1-minute intervals. Data are reported based on digital integration mode (or proportional integration mode), which we have shown corresponds most closely to polysomnography.13 ActionW-2 software (Ambulatory Monitoring Inc) was used to analyze actigraphy data.14 Participants also completed a sleep diary in which they were asked to record bedtime and wake time, known naps, and intervals during which the actigraph was removed (eg, for showering or bathing). This information was used in scoring as previously described.15 Variables used in this analysis included the following: (1) total sleep time (the hours per night spent sleeping while in bed); (2) sleep efficiency (the percentage of time in bed spent sleeping); (3) wake time after sleep onset (minutes of wake after sleep onset during the in-bed intervalsleep onset was defined as completion of 20 continuous minutes of sleep after getting into bed); and (4) daytime inactivity (number of minutes of inactivity during out-of-bed interval, scored as daytime napping by actigraphy scoring algorithm). For convenience, daytime inactivity is referred to as napping throughout the remainder of the article. All exposure variables from actigraphy reflect average daily experience. In a subset of SOF participants who had concurrent actigraphy and polysomnography (the gold standard) data during

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1769

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

Table 1. Characteristics of 2978 Women by Categories of Total Sleep Time


Total Sleep Time, h Characteristic Age, mean (SD), y BMI, mean (SD) MMSE, mean (SD), score (range, 0-30) Take walks for exercise, No. (%) One or more IADL impairment, No. (%) Depression, GDS score 6, No. (%) Possible dementia, No. (%) African American, No. (%) Currently taking benzodiazepines (any), No. (%) Currently taking short-acting benzodiazepines Currently taking long-acting benzodiazepines Currently taking antidepressants, No. (%) Currently taking nonbenzodiazepine receptor agonists, No. (%) Currently taking antipsychotics, No. (%) Goldberg anxiety score, mean (SD) (range, 0-9) Alcoholic drinks per day in the last 30 days, mean (SD), No. Have trouble sleeping due to having to get up to use the bathroom, No. (%) Not during past mo 1 Time per week 1-2 Times per week 3 Times per week 1 Medical conditions, No. (%) Stroke Diabetes Parkinson disease Dementia COPD Cardiovascular disease Cancer Walk speed, mean (SD), m/s Grip strength, mean (SD), kg Used arms to stand from chair, No. (%) Epworth Sleepiness Scale score 10, No. (%) Napping duration, mean (SD), min 5 (n = 254) 83.48 (4.39) 28.92 (6.17) 27.53 (2.13) 61 (24.30) 159 (63.10) 29 (11.42) 15 (6.36) 46 (18.11) 19 (7.48) 13 (5.12) 6 (2.36) 49 (19.29) 4 (1.57) 4 (1.57) 1.46 (2.26) 0.47 (0.89) 5 to 7 (n = 1415) 83.27 (3.64) 27.3 (4.94) 27.91 (1.90) 544 (39.08) 721 (51.32) 152 (10.77) 44 (3.25) 168 (11.87) 97 (6.86) 66 (4.67) 31 (2.19) 156 (11.04) 16 (1.13) 8 (0.57) 1.37 (2.22) 0.50 (0.68) 7 to 8 (n = 897) 83.57 (3.64) 26.53 (4.78) 27.91 (2.08) 357 (40.25) 440 (49.11) 98 (10.95) 25 (2.92) 71 (7.92) 62 (6.92) 41 (4.58) 21 (2.34) 104 (11.61) 7 (0.78) 5 (0.56) 1.25 (2.17) 0.52 (0.73) 8 (n = 412) 84.37 (3.98) 26.03 (4.49) 27.85 (1.92) 139 (33.99) 230 (56.23) 64 (15.57) 18 (4.63) 33 (8.01) 36 (8.74) 30 (7.28) 6 (1.46) 91 (22.09) 3 (0.73) 11 (2.67) 1.37 (2.28) 0.47 (0.66) P Value a .001 .001 .02 .001 .001 .054 .047 .001 .61 .16 .76 .001 .61 .001 .23 .15

42 (16.54) 9 (3.54) 33 (12.99) 170 (66.93) 179 (70.47) 45 (17.72) 42 (16.54) 1 (0.39) 10 (3.94) 49 (19.29) 104 (40.94) 47 (18.50) 0.76 (0.20) 16.68 (4.09) 11 (7.10) 48 (18.90) 96.73 (76.01)

197 (13.94) 113 (8.00) 181 (12.81) 922 (65.25) 877 (62.11) 161 (11.39) 163 (11.54) 12 (0.85) 26 (1.84) 171 (12.10) 458 (32.41) 332 (25.53) 0.84 (0.22) 16.64 (4.13) 50 (4.73) 176 (12.46) 74.08 (59.82)

121 (13.52) 56 (6.26) 103 (11.51) 615 (68.72) 508 (56.76) 110 (12.29) 79 (8.83) 5 (0.56) 12 (1.34) 95 (10.61) 267 (29.83) 181 (20.22) 0.86 (0.21) 16.71 (4.07) 22 (3.34) 65 (7.26) 68.58 (63.81)

52 (12.62) 25 (6.07) 48 (11.65) 287 (69.66) 264 (64.08) 64 (15.53) 43 (10.44) 5 (1.21) 13 (3.16) 42 (10.19) 144 (34.95) 86 (20.87) 0.82 (0.25) 16.16 (4.23) 15 (5.26) 30 (7.28) 79.86 (71.41)

.22

.001 .01 .005 .54 .02 .001 .007 .13 .001 .16 .18 .001 .001

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); COPD, chronic obstructive pulmonary disease; GDS, Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living; MMSE, Mini-Mental State Examination. a P values for continuous variables are from analysis of variance if normally distributed and Kruskal-Wallis test if skewed. P values for categorical variables are from a 2 test.

how often their sleep at night was disturbed by having to get up to go to the bathroom.

STATISTICAL ANALYSIS
Characteristics of participants were compared across categories of total sleep time and by the recurrent falls outcome using 2 tests for categorical variables, analysis of variance for normally distributed continuous variables, and Kruskall Wallis test for skewed continuous variables. Logistic regression was used to assess the association of sleep parameters and subsequent risk of recurrent falls, and results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Characteristics of sleep were analyzed as categorical exposure variables based on current beliefs about clinically relevant values of sleep parameters in older adults and the availability of sufficient numbers of participants in each category. Similar conclusions were obtained when sleep exposure variables were analyzed as continuous exposures (data not shown).

A variety of potential confounders were screened (based on known association with falls) including age, body mass index, medical conditions, walking for exercise, alcohol use, depression, cognitive impairment and possible dementia, frequent bathroom trips during the night, and use of medications including benzodiazepines, antidepressants, and antipsychotics. Variables were considered for inclusion in multivariate models if they were related to total sleep time with a P value of less than .10 or if there was previous literature suggesting an association with sleep patterns. These potential confounders were added to models, along with age and race. Final models included separate dummy variables for shortand long-acting benzodiazepine use vs none; however, results were similar when a single variable (any benzodiazepine use) was included instead. Additional potential mechanisms for the relationship of actigraphic sleep measures and risk of falls were explored by further adjustment for physical performance and subjective daytime sleepiness. All significance levels reported were 2 sided, and all analyses were conducted using SAS version 9.1 (SAS Institute Inc, Cary, North Carolina).
WWW.ARCHINTERNMED.COM

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1770

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

RESULTS

70 60

The mean (SD) age of the analysis cohort was 83.5 (3.8) years, and 10.7% were African American. Mean (SD) total sleep time was 6.8 (1.3) hours, while mean (SD) sleep efficiency was 77.3% (11.7%). On average, participants spent 77.2 (47.4) minutes awake after initial sleep onset. Characteristics of the cohort by categories of total sleep time are given in Table 1. Shorter sleep durations ( 5 hours per night) were associated with decreased cognitive function and possible dementia, slower walking speed, higher body mass index, higher prevalence of chronic obstructive pulmonary disease and cardiovascular disease, and increased subjective daytime sleepiness and napping. Several characteristics appeared to have U-shaped relationships with total sleep time: IADL impairments, prevalence of stroke, antidepressant use, and antipsychotic use were elevated among both short-duration ( 5 hours) and long-duration ( 8 hours) sleepers compared with those with normal total sleep time ( 7 to 8 hours), whereas walking for exercise was less common among short- and long-duration sleepers compared with those with more normal sleep durations. African American women were considerably more likely to be shortduration sleepers (comprising 18.1% of those with total sleep time of 5 hours or less and only 8% of those with total sleep time of more than 8 hours per night) compared with white women. The distribution of number of falls during approximately 1 year after collection of sleep measures is shown in Figure 1. Number of falls ranged from 0 to 38, with a mean (SD) of 0.84 (1.92). A total of 549 women (18.4%) had 2 or more falls during the year after the sleep assessments. As given in Table 2, frequent fallers were older and had slower walk speed; were more likely to have IADL impairments, medical conditions, possible dementia, and lower Mini-Mental State Examination scores; were more likely to use benzodiazepines or antidepressants; and were more likely to report daytime sleepiness. In addition, fallers were more likely to have either short ( 5 hours) or long ( 8 hours) total sleep time and more likely to have fragmented sleep as measured by sleep efficiency and wake time after sleep onset. CHARACTERISTICS OF SLEEP AND RISK OF FALLS After adjustment for age and race, there was a U-shaped pattern of association observed between total sleep time and risk of falls (Table 3). Relative to those with normal sleep ( 7 to 8 hours per night), in analyses adjusting for age and race only, the risk of falls was elevated for those who slept 5 hours or less (OR, 1.82; 95% CI, 1.28-2.57) as well as those who slept more than 8 hours per night (OR, 1.58; 95% CI, 1.17-2.13). The relationship of short sleep ( 5 hours per night) and risk of falls was attenuated but remained significant after multivariate adjustment (OR, 1.52; 95% CI, 1.03-2.24), whereas the relationship of long sleep ( 8 hours per night) and risk of falls was no longer statistically significant (OR, 1.30; 95% CI, 0.94-1.81).

Women Reporting Falls, %

50 40 30 20 10 0 0 1 2 3 4 5 6 7 8

<9

No. of Falls Within the First Year

Figure 1. Percentage of women reporting falls within 1 year after the eighth examination.

Indexes of sleep fragmentation including poor sleep efficiency and greater wake time after sleep onset were associated with increased risk of recurrent falls in models adjusted for age and race and multivariate models (Table 3). In multivariate models, compared with those with sleep efficiency of 70% or higher, those with sleep efficiency lower than 70% had a 1.36fold increased odds of falling (OR, 1.36; 95% CI, 1.071.74). Similarly, those with wake time after sleep onset of 120 minutes of more (compared with 120 minutes) had a 1.33-fold increased odds of falling (OR, 1.32; 95% CI, 1.01-1.71). There was no significant association between napping and risk of falls (Table 3). Furthermore, the interaction of napping and total sleep time was not significant. SLEEP CHARACTERISTICS, USE OF BENZODIAZEPINES, AND RISK OF FALLS In all, 214 subjects (7.2%) reported current use of benzodiazepines. Sixty-four participants reported using longacting benzodiazepines, including clonazepam (n=26), diazepam (n = 24), clorazepate (n = 8), flurazepam hydrochloride (n=3), and chlordiazepoxide (n=3), and 156 participants reported using short-acting benzodiazepines, including lorazepam (n=63), alprazolam (n=49), temazepam (n = 37), triazolam (n = 10), and oxazepam (n=2). Multivariate results given in Table 3 indicate that total sleep time and sleep fragmentation are related to risk of falls, independent of benzodiazepine use. Furthermore, results are similar when benzodiazepine users are excluded from models (data not shown). An elevated but nonsignificant risk of falls associated with the use of benzodiazepines was observed in multivariate models after accounting for total sleep time (Figure 2). Use of any benzodiazepine (short and long combined) was associated with a 1.34-fold increase in risk of falls (OR, 1.34; 95% CI, 0.95-1.90), whereas short- and long-acting benzodiazepine use was associated with an increased odds of 1.43 (95% CI, 0.95-2.15) and 1.18 (95% CI, 0.642.17), respectively. There was no significant interaction
WWW.ARCHINTERNMED.COM

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1771

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

Table 2. Characteristics of 2978 Women by Number of Falls


Falls During the Following Year After Examination 8, No. Characteristic Age, mean (SD), y BMI, mean (SD) MMSE, mean (SD), score (range, 0-30) Take walks for exercise, No. (%) One or more IADL impairment, No. (%) Depression, GDS score 6, No. (%) Possible dementia, No. (%) African American, No. (%) Currently taking benzodiazepines (any), No. (%) Currently taking short-acting benzodiazepines, No. (%) Currently taking long-acting benzodiazepines, No. (%) Currently taking antidepressants, No. (%) Currently taking nonbenzodiazepine receptor agonists, No. (%) Currently taking antipsychotics, No. (%) Goldberg anxiety score, mean (SD) (range, 0-9) Alcoholic drinks per day in the last 30 days, mean (SD), No. Have trouble sleeping due to having to get up to use the bathroom, No. (%) Not during past month 1 Time per week 1-2 Times per week 3 Times per week 1 Medical conditions, No. (%) Stroke Diabetes Parkinson disease COPD Cardiovascular disease Cancer Walk speed, mean (SD), m/s Grip strength, mean (SD), kg Used arms to stand from chair, No. (%) Epworth Sleepiness Scale score 10, No. (%) Napping duration, mean (SD), min Total sleep time categories, No. (%) 5h 5-7 h 7-8 h 8h Sleep efficiency 70%, No. (%) Wake after sleep onset 120 min, No. (%) Napping duration 120 min, No. (%) 0 or 1 (n = 2429) 83.38 (3.79) 27.08 (5.01) 27.92 (1.95) 914 (38.08) 1194 (49.42) 238 (9.81) 72 (3.11) 274 (11.28) 156 (6.43) 110 (4.53) 46 (1.89) 270 (11.12) 22 (0.91) 19 (0.78) 1.26 (2.14) 0.52 (0.73) 331 (13.63) 169 (6.96) 306 (12.60) 1622 (66.80) 1430 (58.92) 279 (11.49) 265 (10.91) 15 (0.62) 282 (11.61) 757 (31.18) 515 (21.22) 0.85 (0.22) 16.78 (4.12) 84 (4.62) 245 (10.09) 73.23 (61.5) 195 (8.03) 1146 (47.18) 766 (31.54) 322 (13.26) 467 (19.23) 357 (14.70) 440 (18.41) 2 (n = 549) 84.17 (4.01) 26.80 (4.95) 27.64 (2.10) 187 (34.69) 356 (65.20) 105 (19.27) 30 (5.84) 44 (8.01) 58 (10.60) 40 (7.31) 18 (3.29) 130 (23.77) 8 (1.46) 9 (1.65) 1.7 (2.44) 0.43 (0.62) 81 (14.84) 34 (6.23) 59 (10.81) 372 (68.13) 375 (68.81) 101 (18.50) 62 (11.36) 8 (1.47) 75 (13.74) 216 (39.56) 131 (24.04) 0.79 (0.22) 15.78 (4.04) 14 (4.18) 74 (13.55) 83.58 (76.39) 59 (10.75) 269 (49.00) 131 (23.86) 90 (16.39) 146 (26.59) 118 (21.49) 118 (21.93) P Value a .001 .25 .004 .14 .001 .001 .003 .03 .001 .007 .04 .001 .24 .06 .001 .02 .55

.001 .001 .77 .055 .17 .001 .15 .001 .001 .72 .02 .04 .001

.001 .06

Abbreviations: BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); COPD, chronic obstructive pulmonary disease; GDS, Geriatric Depression Scale; IADL, Instrumental Activities of Daily Living; MMSE, Mini-Mental State Examination. a P values for continuous variables are from analysis of variance if normally distributed and Kruskal-Wallis test if skewed. P values for categorical variables are from a 2 test.

of benzodiazepine use and sleep duration (or other sleep exposures) in relation to risk of falls. There were too few users of benzodiazepine receptor agonists (n=30) to evaluate the association of this class of sleep medication and risk of falls, and the unadjusted association was not significant (Table 2). ADDITIONAL ANALYSES Of the physical performance measures collected in SOF, the measure of walk speed was most strongly related to total sleep time (Table 1). In general, the magnitude of associations between sleep measures and risk of falls was reduced somewhat by the addition of walk speed to

the models, suggesting that decreased physical performance may partly explain the association of poor sleep and risk of falls. The association of short sleep ( 5 hours per night) and risk of falls was also attenuated by adjustment for walk speed (OR, 1.40; 95% CI, 0.922.12). For the comparison of women with sleep efficiency lower than 70% with others, the odds of falling was decreased from 1.36 in the multivariate model, to 1.17 (95% CI, 0.89-1.52) after the addition of walk speed as covariate. Similarly, the multivariate odds of falling associated with wake time after sleep onset of 120 minutes or more decreased from 1.32 to 1.12 (95% CI, 0.84-1.50) after further adjustment for walk speed.
WWW.ARCHINTERNMED.COM

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1772

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

Table 3. Sleep Characteristics and Risk of Recurrent a Falls During 1 Year After Examination 8
Odds Ratio (95% Confidence Interval) Sleep Characteristic Total sleep time, h 5 5-7 7-8 8 Sleep efficiency, % 70 70 Wake time after sleep onset, min 120 120 Napping duration, min 120 120 No. (%) 254 (8.5) 1415 (47.5) 897 (30.1) 412 (13.8) 2365 (79.4) 613 (20.6) 2502 (84.0) 475 (16.0) 2370 (80.9) 558 (19.1) Age and Race Adjusted 1.82 (1.28-2.57) c 1.41 (1.12-1.77) c 1 [Reference] 1.58 (1.17-2.13) c 1.57 (1.26-1.96) c 1 [Reference] 1 [Reference] 1.58 (1.25-2.00) c 1 [Reference] 1.22 (0.97-1.53) Multivariate Adjusted b 1.52 (1.03-2.24) c 1.40 (1.10-1.79) c 1 [Reference] 1.30 (0.94-1.81) 1.36 (1.07-1.74) c 1 [Reference] 1 [Reference] 1.32 (1.01-1.71) c 1 [Reference] 1.11 (0.86-1.44) Multivariate Adjusted b Walk Speed 1.40 (0.92-2.12) 1.38 (1.07-1.78) c 1 [Reference] 1.26 (0.90-1.78) 1.17 (0.89-1.52) 1 [Reference] 1 [Reference] 1.12 (0.84-1.50) 1 [Reference] 0.99 (0.75-1.30) Multivariate Adjusted b Epworth Score 1.47 (1.00-2.17) 1.38 (1.08-1.76) c 1 [Reference] 1.31 (0.94-1.82) 1.35 (1.06-1.72) c 1 [Reference] 1 [Reference] 1.32 (1.01-1.71) 1 [Reference] 1.11 (0.86-1.43)

a Recurrent falls are defined as 2 or more falls during the subsequent year after the eighth examination. b Adjusted for age, race, body mass index, depression, exercise, instrumental activities of daily living, medical

conditions, possible dementia, and use of

benzodiazepines, antidepressants, and antipsychotics. c P .05.

Odds Ratio (95% Confidence Interval)

Further adjustment of multivariate models for selfreported daytime sleepiness did not substantially affect the results (Table 3).
COMMENT

2.5

2.0

1.5

To our knowledge, this is the first study to demonstrate an association of objective estimates of sleep-wake patterns and risk of recurrent falls in a large communitydwelling population of older women. Estimates of short and more fragmented sleep are strongly related to risk of falls, independent of potential confounders such as age, use of benzodiazepines and other medications, body mass index, and comorbidities. These associations are attenuated somewhat by further adjustment for walk speed. Recently, published data from the SOF have shown that measures of poor physical performance are associated with poor sleep as measured by actigraphy.24 Walk speed may be on the causal pathway between sleep disturbance and risk of falls. Alternatively, short sleep may be a marker for physical frailty, which is related to both poor physical performance and risk of falls. After adjustment for objectively measured total sleep time, use of benzodiazepines was no longer significantly associated with risk of falls, although there remained a suggestive trend in particular for use of short-acting benzodiazepines (OR, 1.43; 95% CI, 0.95-2.15). These findings suggest that the association between the use of these medications and risk of falls among older adults may be in part explained by poor sleep, a possibility which has never been explored using objective measures of sleep. Prior studies of sleep and risk of falls all relied on selfreport of sleep duration and fragmentation, which may be more prone to misclassification of the exposure,25 particularly in older subjects who may be particularly susceptible to cognitive impairments and memory loss. Within our SOF cohort, the correlation between self-

1.0

0.5

0 Any use Short acting Long acting

Use of Benzodiazepines

Figure 2. Association of benzodiazepine use and risk of 2 or more falls during 1 year after the eighth examination, adjusting for age, race, body mass index, exercise, instrumental activities of daily living, medical conditions, possible dementia, use of antidepressants, use of antipsychotics, and total sleep time.

reported nighttime sleep duration and total sleep time based on actigraphy was only 0.26. Using information from a prior SOF visit on selfreported sleep and nap habits, our group recently found that daily napping was associated with risk of falls, whereas nighttime sleep duration was not related to risk of falls.4 In contrast, the present study demonstrates strong associations of actigraphic estimates of short sleep and risk of falls, but no relationship of napping and risk of falls. The divergence in the findings may reflect diminished ability of older subjects to accurately report sleep duration. Furthermore, napping information based on actigraphy includes both intentional and unintentional naps and possibly periods of wake associated with very minimal levels of activity, which may be quite a different kind of exposure than self-reported daily napping.
WWW.ARCHINTERNMED.COM

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1773

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

Most prior studies that have examined sleep as a risk factor for falls have been retrospective, examining selfreported sleep habits in association with history of falls during the previous 12-month interval. Teo et al26 examined the independent effects of self-reported sleep characteristics and urinary incontinence on risk of falls in a population of 786 community-dwelling Australian women who were 75 years and older. Abnormal daytime sleepiness was associated with a 2-fold increase in risk of falls after controlling for fall-related risk factors including urinary incontinence and benzodiazepine use. Similarly, in a population-based survey of older adults (971 women and 555 men), Brassington et al5 found that frequent napping was associated with a nearly 2-fold risk of falls after accounting for a variety of factors including the use of medications, mobility limitations, depression, and comorbidities. In contrast, there was no relationship between self-reported sleep duration and risk of falls. In a study of 34 163 elderly nursing home patients, Avidan and colleagues6 found that untreated insomnia and hypnotic-treated, nonresponsive insomnia were both associated with greater risk of falls (55% and 32% greater risk, respectively); however, hypnotic use alone was not associated with greater risk of falls. These findings are somewhat consistent with results of our study, which indicate that the use of benzodiazepines is not as strongly related to risk of falls after accounting for poor sleep. The association between poor sleep and risk of falls could be mediated through a variety of mechanisms, including impaired cognitive function or depression, balance or gait problems secondary to medical conditions such as postural hypotension, use of medications, or other factors. Results of our multivariate regression models suggest that together these factors may explain some but not all of the relationship between short or more fragmented sleep and risk of falls. Sleep deprivation in younger adults leads to slower reaction times,27 and this may represent an unmeasured factor that could explain these findings. This study has many strengths, including the large sample size of community-dwelling older women, objective estimates of sleep and daytime inactivity patterns, and prospective ascertainment of falls. However, there are also some limitations. Results are not generalizable to other population groups such as institutionalized elderly persons, men, or younger adults. Short or fragmented sleep could be reflective of specific sleep disorders (eg, sleep-related breathing disorder) that cannot be discerned based on actigraphy. Actigraphy is a measure of activity and inactivity and not a definitive characterization of sleep-wake status. As such, actigraphy cannot reliably separate daytime napping from periods of extreme inactivity. Actigraphs were worn for a relatively short period (mean, 3.6 consecutive 24-hour periods), which may affect precision of the estimates. However, any decrease in precision would tend to bias associations toward the null and is offset by the large sample size in our study. Finally, the timing and circumstances of falling were not collected in this study. In conclusion, actigraphic estimates of sleep duration and fragmentation are associated with greater risk of recurrent falls among older women, independent of

use of hypnotic medications commonly used to treat insomnia such as benzodiazepines. Future studies, in particular randomized trials, are needed to determine the effects of newer pharmaceutical interventions for insomnia (eg, benzodiazepine receptor agonists) or cognitive behavioral therapy for insomnia on risk of falls. In addition, future studies using comprehensive and objective measures of sleep should examine the interrelationships between specific sleep characteristics (eg, sleeprelated breathing disorder, hypoxia, and measures of sleep duration and fragmentation) to determine if these disorders contribute independently toward risk of falls. Accepted for Publication: February 26, 2008. Author Affiliations: Research Institute, California Pacific Medical Center and San Francisco Coordinating Center, San Francisco (Drs Stone and Cummings and Mss Blackwell and Schneider); Department of Psychiatry, University of California, San Diego (Dr Ancoli-Israel); Center for Chronic Disease Outcomes Research, Veterans Affairs Medical Center, and Department of Medicine and Division of Epidemiology and Community Health, University of Minnesota, Minneapolis (Dr Ensrud); Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, Pennsylvania (Dr Cauley); Center for Clinical Investigation and Department of Medicine, Case Western Reserve University, Cleveland, Ohio (Dr Redline); Kaiser Permanente Center for Health Research Northwest and Hawaii, Portland, Oregon (Dr Hillier); and Departments of Medicine (Dr Claman) and Epidemiology and Biostatistics (Dr Cummings), University of California, San Francisco. Correspondence: Katie L. Stone, PhD, San Francisco Coordinating Center, California Pacific Medical Center Research Institute, 185 Berry St, Lobby 4, Fifth Floor, Ste 5700, San Francisco, CA 94107-1762 (kstone@sfcc-cpmc .net). Author Contributions: Study concept and design: Stone, Ancoli-Israel, Redline, and Cummings. Acquisition of data: Stone, Ancoli-Israel, Ensrud, Cauley, Redline, Hillier, and Schneider. Analysis and interpretation of data: Stone, Ancoli-Israel, Blackwell, Ensrud, Cauley, Redline, Hillier, and Claman. Drafting of the manuscript: Stone and AncoliIsrael. Critical revision of the manuscript for important intellectual content: Ancoli-Israel, Blackwell, Ensrud, Cauley, Redline, Hillier, Schneider, Claman, and Cummings. Statistical analysis: Stone and Blackwell. Obtained funding: Stone, Ensrud, Cauley, Hillier, and Cummings. Administrative, technical, and material support: Redline, Schneider, and Cummings. Study supervision: Stone and AncoliIsrael. Financial Disclosure: Dr Stone is a consultant for Sepracor Inc and an invited speaker for Sanofi-Aventis. Dr Ancoli-Israel is a consultant and on the scientific advisory board for Arena, Cephalon Inc, Neurocrine Biosciences Inc, Sanofi-Aventis, Sepracor Inc, and Takeda Pharmaceuticals North America Inc and has received grant support from Sepracor Inc and Takeda Pharmaceuticals North America Inc. Funding/Support: This study was supported by Public Health Service grants AG05407, AR35582, AG05394, AR35584, AR35583, and AG08415.
WWW.ARCHINTERNMED.COM

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1774

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

Previous Presentation: This work was presented in part as an oral presentation at the Associated Professional Sleep Societies Annual Meeting; June 9, 2004; Philadelphia, Pennsylvania.
REFERENCES
1. Nevitt MC, Cummings SR, Hudes ES. Risk factors for injurious falls: a prospective study. J Gerontol. 1991;46(5):M164-M170. 2. Nevitt MC, Cummings SR, Kidd S, Black D. Risk factors for recurrent nonsyncopal falls: a prospective study. JAMA. 1989;261(18):2663-2668. 3. OLoughlin JL, Robitaille Y, Boivin JF, Suissa S. Incidence of and risk factors for falls and injurious falls among the community-dwelling elderly. Am J Epidemiol. 1993;137(3):342-354. 4. Stone KL, Ewing SK, Lui LY, et al. Self-reported sleep and nap habits and risk of falls and fractures in older women: the study of osteoporotic fractures. J Am Geriatr Soc. 2006;54(8):1177-1183. 5. Brassington GS, King AC, Bliwise DL. Sleep problems as a risk factor for falls in a sample of community-dwelling adults aged 64-99 years. J Am Geriatr Soc. 2000; 48(10):1234-1240. 6. Avidan AY, Fries BE, James ML, Szafara KL, Wright GT, Chervin RD. Insomnia and hypnotic use, recorded in the minimum data set, as predictors of falls and hip fractures in Michigan nursing homes. J Am Geriatr Soc. 2005;53(6):955962. 7. Ensrud KE, Blackwell TL, Mangione CM, et al. Central nervous system-active medications and risk for falls in older women. J Am Geriatr Soc. 2002;50(10):16291637. 8. Ensrud KE, Blackwell T, Mangione CM, et al. Central nervous system active medications and risk for fractures in older women. Arch Intern Med. 2003;163(8): 949-957. 9. Allain H, Bentue-Ferrer D, Polard E, Akwa Y, Patat A. Postural instability and consequent falls and hip fractures associated with use of hypnotics in the elderly: a comparative review. Drugs Aging. 2005;22(9):749-765. 10. Landi F, Onder G, Cesari M, Barillaro C, Russo A, Bernabei R. Psychotropic medications and risk for falls among community-dwelling frail older people: an observational study. J Gerontol A Biol Sci Med Sci. 2005;60(5):622-626. 11. Stenbacka M, Jansson B, Leifman A, Romelsjo A. Association between use of sedatives or hypnotics, alcohol consumption, or other risk factors and a single injurious fall or multiple injurious falls: a longitudinal general population study. Alcohol. 2002;28(1):9-16. 12. Cummings SR, Nevitt MC, Browner WS, et al; Study of Osteoporotic Fractures

13.

14. 15.

16.

17.

18. 19. 20.

21.

22. 23. 24.

25. 26.

27.

Research Group. Risk factors for hip fracture in white women. N Engl J Med. 1995;332(12):767-773. Blackwell T, Redline S, Ancoli-Israel S, et al. Comparison of sleep parameters from actigraphy and polysomnography in older women: the SOF study. Sleep. 2008;31(2):283-291. Action-W Users Guide, Version 2.0. Ardsley, NY: Ambulatory Monitoring Inc; 1999. Blackwell T, Ancoli-Israel S, Gehrman PR, Schneider JL, Pedula KL, Stone KL. Actigraphy scoring reliability in the study of osteoporotic fractures. Sleep. 2005; 28(12):1599-1605. Pahor M, Chrischilles EA, Guralnik JM, Brown SL, Wallace RB, Carbonin P. Drug data coding and analysis in epidemiologic studies. Eur J Epidemiol. 1994; 10(4):405-411. Gerety MB, Williams JW Jr, Mulrow CD, et al. Performance of case-finding tools for depression in the nursing home: influence of clinical and functional characteristics and selection of optimal threshold scores. J Am Geriatr Soc. 1994; 42(10):1103-1109. Yesavage JA. Geriatric Depression Scale. Psychopharmacol Bull. 1988;24(4):709711. Goldberg D, Bridges K, Duncan-Jones P, Grayson D. Detecting anxiety and depression in general medical settings. BMJ. 1988;297(6653):897-899. Crum RM, Anthony JC, Bassett SS, Folstein MF. Population-based norms for the Mini-Mental State Examination by age and educational level. JAMA. 1993;269 (18):2386-2391. Nygard L. Instrumental activities of daily living: a stepping-stone towards Alz heimers disease diagnosis in subjects with mild cognitive impairment? Acta Neurol Scand Suppl. 2003;179:42-46. Johns MW. Reliability and factor analysis of the Epworth Sleepiness Scale. Sleep. 1992;15(4):376-381. Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep. 1991;14(6):540-545. Goldman SE, Stone KL, Ancoli-Israel S, et al. Poor sleep is associated with poorer physical performance and greater functional limitations in older women. Sleep. 2007;30(10):1317-1324. Means MK, Edinger JD, Glenn DM, Fins AI. Accuracy of sleep perceptions among insomnia sufferers and normal sleepers. Sleep Med. 2003;4(4):285-296. Teo JS, Briffa NK, Devine A, Dhaliwal SS, Prince RL. Do sleep problems or urinary incontinence predict falls in elderly women? Aust J Physiother. 2006; 52(1):19-24. Doran SM, Van Dongen HP, Dinges DF. Sustained attention performance during sleep deprivation: evidence of state instability. Arch Ital Biol. 2001;139(3):253267.

(REPRINTED) ARCH INTERN MED/ VOL 168 (NO. 16), SEP 8, 2008 1775

WWW.ARCHINTERNMED.COM

Downloaded from www.archinternmed.com at Capes Consortia, on November 29, 2009 2008 American Medical Association. All rights reserved.

You might also like