To compare sleep and 24-hour rest/activity rhythms (RARs) between cognitively normal older adults who are β-amyloid-positive (Aβ+) or Aβ- and replicate a novel time-of-day-specific difference between these groups identified in a previous exploratory study.
Digital health technologies (DHTs) enable us to measure human physiology and behavior remotely, objectively and continuously. With the accelerated adoption of DHTs in clinical trials, there is an unmet need to identify statistical approaches to address missing data to ensure that the derived endpoints are valid, accurate, and reliable. It is not obvious how commonly used statistical methods to handle missing data in clinical trials can be directly applied to the complex data collected by DHTs. Meanwhile, current approaches used to address missing data from DHTs are of limited sophistication and focus on the exclusion of data where the quantity of missing data exceeds a given threshold. High-frequency time series data collected by DHTs are often summarized to derive epoch-level data, which are then processed to compute daily summary measures. In this article, we discuss characteristics of missing data collected by DHT, review emerging statistical approaches for addressing missingness in epoch-level data including within-patient imputations across common time periods, functional data analysis, and deep learning methods, as well as imputation approaches and robust modeling appropriate for handling missing data in daily summary measures. We discuss strategies for minimizing missing data by optimizing DHT deployment and by including the patients' perspectives in the study design. We believe that these approaches provide more insight into preventing missing data when deriving digital endpoints. We hope this article can serve as a starting point for further discussion among clinical trial stakeholders.
Survey satisficing occurs when participants respond to survey questions rapidly without carefully reading or comprehending them. Studies have demonstrated the occurrence of survey satisficing, which can degrade survey quality, particularly in longitudinal studies.The aim of this study is to use a group-based trajectory analysis method to identify satisficers when similar survey questions were asked periodically in a long-standing cohort, and to examine factors associated with satisficing in the surveys having sensitive human immunodeficiency virus (HIV)-related behavioral questions.Behavioral data were collected semiannually online at all four sites of the Multicenter AIDS Cohort Study (MACS) from October 2008 through March 2013. Based on the start and end times, and the word counts per variable, response speed (word counts per second) for each participant visit was calculated. Two-step group-based trajectory analyses of the response speed across 9 study visits were performed to identify potential survey satisficing. Generalized linear models with repeated measures were used to investigate the factors associated with satisficing on HIV-related behavioral surveys.Among the total 2138 male participants, the median baseline age was 51 years (interquartile range, 45-58); most of the participants were non-Hispanic white (62.72%, 1341/2138) and college graduates (46.59%, 996/2138), and half were HIV seropositive (50.00%, 1069/2138). A total of 543 men (25.40%, 543/2138) were considered potential satisficers with respect to their increased trajectory tendency of response speed. In the multivariate analysis, being 10 years older at the baseline visit increased the odds of satisficing by 44% (OR 1.44, 95% CI 1.27-1.62, P<.001). Compared with the non-Hispanic white participants, non-Hispanic black participants were 122% more likely to satisfice the HIV-related behavioral survey (OR 2.22, 95% CI 1.69-2.91, P<.001), and 99% more likely to do so for the other race/ethnicity group (OR 1.99, 95% CI 1.39-2.83, P<.001). Participants with a high school degree or less were 67% more likely to satisfice the survey (OR 1.67, 95% CI 1.26-2.21, P<.001) compared with those with a college degree. Having more than one sex partner and using more than one recreational drug reduced the odds of satisficing by 24% (OR 0.76, 95% CI 0.61-0.94, P=.013) and 28% (OR 0.72, 95% CI 0.55-0.93, P=.013), respectively. No statistically significant association of HIV serostatus with satisficing was observed.Using a group-based trajectory analysis method, we could identify consistent satisficing on HIV-related behavioral surveys among participants in the MACS, which was associated with being older, being non-white, and having a lower education level; however, there was no significant difference by HIV serostatus. Methods to minimize satisficing using longitudinal survey data are warranted.
Importance Accelerometry has been increasingly used as an objective index of sleep, physical activity, and circadian rhythms in people with mood disorders. However, most prior research has focused on sleep or physical activity alone without consideration of the strong within- and cross-domain intercorrelations; and few studies have distinguished between trait and state profiles of accelerometry domains in major depressive disorder (MDD). Objectives To identify joint and individual components of the domains derived from accelerometry, including sleep, physical activity, and circadian rhythmicity using the Joint and Individual Variation Explained method (JIVE), a novel multimodal integrative dimension-reduction technique; and to examine associations between joint and individual components with current and remitted MDD. Design, Setting, and Participants This cross-sectional study examined data from the second wave of a population cohort study from Lausanne, Switzerland. Participants included 2317 adults (1164 without MDD, 185 with current MDD, and 968 with remitted MDD) with accelerometry for at least 7 days. Statistical analysis was conducted from January 2021 to June 2023. Main Outcomes and Measures Features derived from accelerometry for 14 days; current and remitted MDD. Logistic regression adjusted for age, sex, body mass index, and anxiety and substance use disorders. Results Among 2317 adults included in the study, 1261 (54.42%) were female, and mean (SD) age was 61.79 (9.97) years. JIVE reduced 28 accelerometry features to 3 joint and 6 individual components (1 sleep, 2 physical activity, 3 circadian rhythms). Joint components explained 58.5%, 79.5%, 54.5% of the total variation in sleep, physical activity, and circadian rhythm domains, respectively. Both current and remitted depression were associated with the first 2 joint components that were distinguished by the salience of high-intensity physical activity and amplitude of circadian rhythm and timing of both sleep and physical activity, respectively. MDD had significantly weaker circadian rhythmicity. Conclusions and Relevance Application of a novel multimodal dimension-reduction technique demonstrates the importance of joint influences of physical activity, circadian rhythms, and timing of both sleep and physical activity with MDD; dampened circadian rhythmicity may constitute a trait marker for MDD. This work illustrates the value of accelerometry as a potential biomarker for subtypes of depression and highlights the importance of consideration of the full 24-hour sleep-wake cycle in future studies.
Abstract Introduction Chronotype is a potentially modifiable contributor to human well-being and longevity, with eveningness commonly linked to poorer outcomes. We examined the relationship between actigraphy-measured chronotype and all-cause mortality in a nationally representative sample of US adults. We also examined the association between social jetlag, a measure of circadian misalignment, and all-cause mortality. Methods Data were from 2,256 participants ≥50 from the National Health and Nutrition Examination Survey 2003-2006 cohorts. Participants were asked to wear a hip-worn Actigraph 7164 uni-axial activity monitor for 7 days, and to remove the device for sleep. Objectively-measured bedtime (OBT) was computed as the start of the non-wear period with the longest duration within each 24h period. Duration of the in-bed period (OBT-D) was computed as the hours from OBT to the end of the in-bed period. Midpoint of OBT (OBT-M) was computed as the midpoint between OBT and the end of the in-bed period. Chronotype was estimated using the average OBT-M separately for weekdays, weekends (Friday and Saturday nights), and all days combined. A weekend OBT-M corrected for sleep debt for participants with weekend OBT-D>weekday OBT-D was also computed. The following formula was applied to correct for sleep debt: weekend OBT-M minus ((weekend OBT-D minus weekday OBT-D)/2). Consistent with previous research, OBT-Ms were categorized into intermediate (≥3:30am & ≤4:30am), morningness (<3:30am), and eveningness (>4:30am) chronotypes. Social jetlag was defined as the difference between weekend and weekday OBT-Ms and expressed in hours. Survey-weighted Cox proportional hazard models were used to examine the relationship between circadian factors and all-cause mortality. There were 642 deaths, excluding accidental deaths. Results Adjusted for age, sex, race, SES, BMI, smoking and drinking status, comorbidities, and average OBT-D, an eveningness chronotype (i.e., weekend OBT-M corrected for sleep debt) was associated with a greater hazard of death compared to an intermediate chronotype (HR=1.68, 95% CI=1.25, 2.26). There were no other significant associations. Conclusion Evening-oriented chronotype is associated with greater mortality risk in adults aged ≥50. To our knowledge, this is the first study to report the link between chronotype, estimated objectively via actigraphy, and all-cause mortality in a nationally representative sample. Support (if any) NIH grant 5T32MH014592-39.
Disturbed sleep may increase AD risk, but much less is known about links between circadian rest/activity rhythms (RARs) and brain health in cognitively normal older adults. We determined associations between data-driven indices of RARs and regional brain volumes, including regions showing early change in Alzheimer's disease (AD), in a cognitively normal sample. We studied 275 participants in the Baltimore Longitudinal Study of Aging who underwent 3-T magnetic resonance imaging of the brain, completed 5.8 ±0.6 24-hour periods of wrist actigraphy, the Mini-Mental State Examination (MMSE), and the Center for Epidemiological Studies Depression Scale (CES-D). Using functional principal component (fPC) analysis, we identified 10 uncorrelated patterns (fPCs) accounting for 90% of variability in actigraphic RARs. Participants' fPC scores were included simultaneously in models as primary predictors. Outcomes were volumes of gray matter, white matter, ventricles, inferior temporal gyrus, hippocampus, and parahippocampal gyrus, all normalized to intracranial volume. We adjusted for age, sex, race, education, CES-D and MMSE. Results were rounded to the nearest 30 minutes. Participants were aged 72.9 ±11.5 years, 56.0% women, and 28.7% racial/ethnic minorities. fPC1 represented the average 24-hour RAR pattern across all subjects, in which activity sharply increased at 6am, peaked at 8:30am, then slowly decreased until 9pm, when it decreased sharply until 6am. Participants with higher fPC1 scores had greater white matter volume (p=0.009) and smaller ventricles (p=0.005). Other patterns were linked to lower brain volumes. For example, higher fPC4 (reflecting higher activity from 4:30-8am and lower activity from 8-12:30pm) and higher fPC8 scores (representing an "ultradian" pattern, with repeated cycles of high-to-low activity within a day) were linked to lower inferior temporal gyrus volume (p<0.008 for both); fPC4 was also linked to lower hippocampal volume (p=0.047). Finally, higher fPC10 scores (representing an ultradian pattern distinct from fPC8) were associated with lower parahippocampal gyrus (p=0.023) and greater ventricular volume (p=0.001). In cognitively normal older adults, atypical RARs are associated with lower brain volumes, including in regions affected early in AD. Further research is needed on RARs as modifiable contributors to, or markers and predictors of preclinical neurodegeneration.
Background: We propose a method for estimating the timing of in-bed intervals using objective data in a large representative US sample, and quantify the association between these intervals and age, sex, and day of the week.Methods: The study included 11,951 participants 6 years and older from the National Health and Nutrition Examination Survey (NHANES) 2003–2006, who wore accelerometers to measure physical activity for seven consecutive days. Participants were instructed to remove the device just before the nighttime sleep period and put it back on immediately after. This nighttime period of non-wear was defined in this paper as the objective bedtime (OBT), an objectively estimated record of the in-bed interval. For each night of the week, we estimated two measures: the duration of the OBT (OBT-D) and, as a measure of the chronotype, the midpoint of the OBT (OBT-M). We estimated day-of-the-week-specific OBT-D and OBT-M using gender-specific population percentile curves. Differences in OBT-M (chronotype) and OBT-D (the amount of time spent in bed) by age and sex were estimated using regression models.Results: The estimates of OBT-M and their differences among age groups were consistent with the estimates of chronotype obtained via self-report in European populations. The average OBT-M varied significantly by age, while OBT-D was less variable with age. In the reference group (females, aged 17–22 years), the average OBT-M across 7 days was 4:19 AM (SD = 30 min) and the average OBT-D was 9 h 19 min (SD = 12 min). In the same age group the average OBT-D was 18 minutes shorter for males than for females, while the average OBT-M was not significantly different between males and females. The most pronounced differences were observed between OBT-M of weekday and weekend nights. In the reference group, compared to the average OBT-M of 3:50 am on Monday through Thursday nights, there was a 57-minute delay in OBT-M on Friday nights (entering the weekend), a 69-minute delay on Saturday nights (staying in the weekend), and a 23-minute delay on Sunday night (leaving the weekend). For both OBT-M and OBT-D, in most age groups and for most days of the week, there were no statistically significant differences between males and females, except for OBT-D on Wednesdays and Thursdays, with males having 31 (p-value < 0.05) and 45 (p-value < 0.05) minutes shorter OBT-D, respectively.Conclusions: The proposed measures, OBT-D and OBT-M, provide useful information of time in bed and chronotype in NHANES 2003–2006. They identify within-week patterns of bedtime and can be used to study associations between the bedtime and the large number of health outcomes collected in NHANES 2003–2006.
Background: We propose a method for estimating the timing of in-bed intervals using objective data in a large representative U.S. sample, and quantify the association between these intervals and age, sex, and day of the week. Methods: The study included 11,951 participants six years and older from the National Health and Nutrition Examination Survey (NHANES) 2003-2006, who wore accelerometers to measure physical activity for seven consecutive days. Participants were instructed to remove the device just before the nighttime sleep period and put it back on immediately after. This nighttime period of non-wear was defined in this paper as the objective bedtime (OBT), an objectively estimated record of the in-bed-interval. For each night of the week, we estimated two measures: the duration of the OBT (OBT-D) and, as a measure of the chronotype, the midpoint of the OBT (OBT-M). We estimated day-of-the-week-specific OBT-D and OBT-M using gender-specific population percentile curves. Differences in OBT-M (chronotype) and OBT-D (the amount of time spent in bed) by age and sex were estimated using regression models. Results: The estimates of OBT-M and their differences among age groups were consistent with the estimates of chronotype obtained via self-report in European populations. The average OBT-M varied significantly by age, while OBT-D was less variable with age. The most pronounced differences were observed between OBT-M of weekday and weekend nights. Conclusions: The proposed measures, OBT-D and OBT-M, provide useful information of time in bed and chronotype in NHANES 2003-2006. They identify within-week patterns of bedtime and can be used to study associations between the bedtime and the large number of health outcomes collected in NHANES 2003-2006.
