Methods
Participants
During the baseline examination of the Osteoporotic Fractures in Men (MrOS) study (2000–2002), 5,994 community-dwelling men aged 65 and older were enrolled at six sites in the United States: Birmingham, Alabama; Minneapolis, Minnesota; Palo Alto, California; the Monon-gahela Valley near Pittsburgh, Pennsylvania; Portland, Oregon; and San Diego, California. The men had to be able to walk unassisted and must not have had a bilateral hip replacement.
The MrOS Sleep Study (2003–2005), an ancillary study of the MrOS Study, recruited 3,135 of these participants (56% of active survivors, >100% of recruitment goal) for a comprehensive sleep assessment; 3,101 of these had data available for incident falls within the first year after the visit.
All men provided written informed consent, and the institutional review board at each site approved the study.
Sleep Parameters
Self-Reported Sleep Parameters. Participants completed the Pittsburgh Sleep Quality Index (PSQI, range 0–21), a validated measure of subjective sleep quality over a 1-month time period. A standard score of >5 is indicative of poor sleep quality.
Participants also completed the Epworth Sleepiness Scale (ESS, range 0–24), a self-administered questionnaire that classifies daytime sleepiness. An ESS score >10 indicates excessive daytime sleepiness.
Participants were asked to self-report total sleep time (sTST), which was categorized as 5 hours or less, >5 to 7 hours, >7 to 8 hours, and more than 8 hours.
Actigraphic Parameters of Sleep–Wake Activity. Actigraphic characteristics of sleep–wake activity were estimated using an actigraph (SleepWatch-O; Ambulatory Monitoring, Inc., Ardsley, NY), a device used to detect movement that is similar in appearance to a wristwatch. Details of the data collection and sleep scoring algorithms used in the study have been published elsewhere.
Participants were instructed to wear the actigraph on the nondominant wrist for a minimum of five nights (5.2 ± 0.9 nights). Participants completed sleep diaries, which were used in editing the data files, as described previously.
Variables used in this analysis included total sleep time (TST; hours spent sleeping after "lights off"), sleep efficiency (an index of sleep fragmentation; percentage of time spent sleeping after "lights off"), and nap time (time scored as sleep for blocks of ≥5 minutes before "lights off"). All variables reflect average daily experience. Variables were categorized as TST of 5 hours or less, 5 to 7 hours, 7 to 8 hours, and more than 8 hours; sleep efficiency as <70% and 70% or greater; and napping as 2 hours or more and <2 hours.
Polysomnographic Parameters of SDB. In-home sleep studies were completed using unattended, portable polysomnography (Safiro; Compumedics, Inc., Melbourne, Vic., Australia). The recording montage included: C3/A2 and C4/A1 electroencephalograms, bilateral electrooculograms and a bipolar submental electromyogram, thoracic and abdominal respiratory inductance plethysmography, airflow (by nasal–oral thermocouple and nasal pressure cannula), finger pulse oximetry, lead I electrocardiogram, body position (mercury switch sensor), and bilateral tibialis leg movements (piezoelectric sensors). Centrally trained, certified staff performed home visits using a protocol similar to that used in the Sleep Heart Health Study. The studies were scored at the Sleep Reading Center (Case Western Reserve University).
Sleep disordered breathing was measured using the apnea hypopnea index (AHI). Nocturnal hypoxemia was defined as percentage of sleep time with arterial oxygen saturation (SaO2) below 90%. Apnea was defined as complete or near-complete cessation of airflow for longer than 10 seconds. Hypopneas were scored if clear reductions in breathing amplitude (≥30% below baseline breathing) occurred and lasted longer than 10 seconds. Only apneas and hypopneas with a desaturation of 3% or greater were included. AHI was calculated as the number of apneas and hypopneas per hour of sleep. Variables were categorized as AHI <5, 5 to 14, 15 to 29, and 30 or more and percentage of sleep time with SaO2 <90% of 10% or more versus <10%.
Ascertainment of Incident Falls
Participants were contacted by postcard or telephone every 4 months to ascertain self-reported incident falls. Response rates exceeded 99%.
Falls reported on the first three postcards returned after the sleep assessment (covering 0.99 ± 0.02 years) were included in this analysis. The outcome for this analysis was recurrent falls, defined as two or more falls (vs ≤1 falls) in the year after the sleep assessment.
Other Measurements
Participants completed questionnaires about demographic characteristics, medical history, physical activity, smoking, and alcohol use. Participants reported how often their sleep was disturbed by having to go to the bathroom. Lower urinary tract symptoms were assessed using the American Urological Association Symptom Index. Medications used within the preceding 30 days were matched to their ingredient(s) based on the Iowa Drug Information Service Drug Vocabulary (College of Pharmacy, University of Iowa, Iowa City, IA). Depression was defined as six or more symptoms on the Geriatric Depression Scale. Anxiety was assessed using the Goldberg Anxiety Scale, activity using the Physical Activity Scale for the Elderly, and cognitive function using the modified Mini-Mental State Examination. Body weight and height were used to calculate body mass index. Functional status was assessed for five instrumental activities of daily living. Physical function was measured according to walking speed (time to walk 6 m at usual pace).
Statistical Analyses
The sleep predictor variables were expressed categorically, defined similarly to previous publications for comparability or using standard cut-points.
Characteristics of participants were compared between categories of actigraphic TST using analysis of variance for normally distributed continuous variables, Kruskal–Wallis tests for skewed continuous variables, and chi-square tests for categorical variables. Similar comparisons were performed between categories of the other sleep predictors (data not shown).
The association between a given sleep parameter and risk of recurrent falls was examined using logistic regression, presented as odds ratios (ORs) with 95% confidence intervals (CIs). Models were minimally adjusted for age, ethnicity, and clinic. Additional covariates were included in a multivariable model if they were related to recurrent falls and one or more sleep parameters in univariate analyses at P < .10.
Secondary analyses were performed for models with SDB predictors removing men from the analyses who reported using continuous positive air pressure at the beginning of the study (n = 55) or during follow-up (n = 64). Multivariable models including the predictors of actigraphic TST and sleep fragmentation, excessive daytime sleepiness, and nocturnal hypoxemia were examined to determine whether associations to recurrent falls were independent.
All significance levels reported were two sided. All analyses were conducted using SAS version 9.2 (SAS Institute, Inc., Cary, NC).