Association Between Sleep Quantity and Quality in Early Adulthood With Cognitive Function in Midlife

Abstract

Background and Objective

Growing evidence supports an association between sleep quality and risk of dementia. However, little is known about whether objectively measured sleep duration and quality influence cognition in midlife, a period of importance for understanding the direction of the association between sleep and dementia. We examined the association between sleep duration and quality, measured when participants were in their mid-30s to late 40s, and midlife cognition assessed 11 years later among Black and White adults.

Methods

As part of the Coronary Artery Risk Development in Young Adults cohort study, sleep duration and quality were assessed objectively using wrist actigraphy and subjectively by Pittsburgh Sleep Quality Index (PSQI) at 2003–2005. During 2015–2016, we evaluated midlife cognition using the Digit Symbol Substitution Test (DSST), Stroop test, Rey Auditory Verbal Learning Test, Montreal Cognitive Assessment (MoCA), and Letter Fluency and Category Fluency tests. We used multivariable logistic regression to examine the association between sleep parameters and poor cognitive performance, which was defined as a score that was >1 SD below the mean score.

Results

The 526 participants (58% women and 44% Black) had a mean age of 40.1 ± 3.6 years at baseline, a mean sleep duration of 6.1 ± 1.1 hours, and mean sleep fragmentation index (calculated as the sum of the percentage of time spent moving and the percentage of immobile periods ≤1 minute) of 19.2 ± 8.1%, and 239 (45.6%) participants reported poor sleep as defined by a PSQI global score of >5. After adjustment for demographics, education, smoking, body mass index, depression, physical activity, hypertension, and diabetes, those in the highest vs lowest tertile of sleep fragmentation index had over twice the odds of having poor cognitive performance (>1 SD below the mean) on the DSST (odds ratio [OR] = 2.97; 95% CI 1.34–6.56), fluency (OR = 2.42; 95% CI 1.17–5.02), and MoCA test (OR = 2.29; 95% CI 1.06–4.94). The association between sleep fragmentation and cognitive performance did not differ by race or sex. Objective sleep duration or subjective sleep quality was not associated with cognition in midlife.

Discussion

Actigraphy-measured high sleep fragmentation rather than sleep duration was associated with worse cognition among middle-aged Black and White men and women. Sleep quality is important for cognitive health even as early as midlife.

Over the past decade, growing evidence suggests an association between sleep and cognitive aging,¹ supporting the role of sleep disturbances as a potential modifiable risk factor for Alzheimer disease (AD).^2,3 However, most previous studies have examined both sleep disturbances and cognitive impairment in late life,^4-6 many with only a few years of follow-up, which raises concerns about “reverse causality.” Given that AD pathology begins to accumulate in the brain several decades before clinical symptoms onset,⁷ it is plausible that sleep disturbances assessed for the first time in late life, close to the onset of clinical AD, might themselves be the consequence of underlying AD pathology that has been developing silently over the years. Understanding of the association between sleep and cognition earlier in life is critical for addressing this concern and understanding the role of sleep disturbances as a risk factor for AD.

Midlife is a crucial period when cognitive decline could begin to manifest and sleep disturbances may differ from those experienced in late life.^8,9 Notably, sleep disturbances in midlife could be driven by multiple life stage–related physiologic processes such as menopause¹⁰ and psychosocial factors, such as work stress,¹¹ which may also affect age-related cognitive decline.^12,13 To date, studies that examined sleep in midlife have primarily focused on cognitive outcomes in late life and relied on subjective measures of sleep or measures of sleep quantity only.^14-17 Subjective measures of sleep are poorly correlated with objective measures and prone to measurement errors.^18,19 A few studies found an association between sleep duration and cognitive performance,^15,17 whereas others suggested that sleep quality rather than quantity is the key.²⁰ There is a paucity of studies of both objective and subjective sleep, both duration and quality, on cognition in midlife. Importantly, both sleep disturbances and cognition vary significantly by race,^21,22 yet little is known about the association between sleep and midlife cognition in ethnically diverse populations. It was hypothesized that the association between sleep and cognition may be stronger among Black adults, given poorer sleeping environments, limited access to health care, and inadequate treatment of diseases (e.g., hypertension and diabetes) that may affect cognitive function.^23,24 Meanwhile, it remains controversial whether sleep influences cognition differently in men vs women because of potential sex differences in a number of factors that may drive the association between sleep and cognition.²⁵

Clarification of the link between different sleep parameters and cognition in midlife offers a valuable opportunity for the early identification of dementia risk and informs the design of future sleep intervention trials, especially in identifying the optimal timing for intervention to reduce dementia risk. In the ongoing Coronary Artery Risk Development in Young Adults (CARDIA) study of Black and White adults, we examined the association between sleep duration and quality, measured both subjectively and objectively by actigraphy when participants were in their mid-30s to late 40s, and midlife cognition assessed approximately 11 years later. We hypothesized that poorer sleep in early midlife will be associated with worse cognition in late midlife. We also explored if the association may differ by race or sex.

Methods

Study Setting and Participants

The CARDIA study is an ongoing multicenter population-based longitudinal cohort of 5,115 Black and White men and women, aged 18–30 years, who were recruited with balance of sex, race, age, and education, and completed baseline assessment during 1985–1986 at 4 clinic sites. Follow-up assessments occurred every 2–5 years over 30 years. At Year 15 follow-up examination, all nonpregnant participants at the Chicago site were invited to take part in the ancillary CARDIA Sleep Study, which took place from 2003 to 2005 and included both subjective and objective assessment of sleep duration and quality. In total, 670 of the 814 eligible participants consented and took part in the sleep examination.²⁶ Our final analytic sample consisted of 526 participants who completed the sleep examination from 2003 to 2005 (between Year 15 and Year 20 examinations) and had cognition evaluated at Year 30 (2015–2016).

Standard Protocol Approvals, Registrations, and Patient Consents

All participants provided written informed consent, with Institutional Review Board approval at Northwestern University and the University of Chicago.

Sleep

Sleep duration and quality were assessed objectively using a wrist activity monitor (Actiwatch-16; Mini-Mitter, Inc., Bend, OR), worn for 3 consecutive days on 2 occasions (waves) approximately 1 year apart. For each occasion, participants were asked to wear the activity monitor from a Wednesday afternoon to a Saturday morning, totaling 6 days/nights over 2 occasions. The averages for all available days were computed and analyzed, given the stability of actigraphy readings across the 2 occasions.¹ The validated manufacturer software (Actiware 3.4, Mini Mitter, Inc) scores each 30-second epoch as sleep or wake based on the activity counts, and sleep onset and end times were derived. We also asked the participants to report bedtimes and wake times in a sleep diary to confirm the time in bed for Actiware sleep detection. This analysis focused on 2 daily sleep measures calculated from actigraphy data: sleep duration, the sum of epochs scored as “sleep” between sleep onset and sleep end, and sleep fragmentation index, a measure of restlessness during the sleep period calculated as the sum of the percentage of time spent moving and the percentage of immobile periods ≤1 minute. Subjective sleep quality was assessed using the well-validated Pittsburgh Sleep Quality Index (PSQI),² with higher scores indicating poorer sleep quality, and a global score of >5 indicative of poor sleep quality. Risk of sleep apnea was determined by the Berlin Questionnaire,⁴ which defines high risk of sleep apnea as the presence of at least 2 of 3 of the following: frequent daytime sleepiness, persistent snoring symptoms, and having obesity or hypertension.

Cognitive Function

At Year 30 (2015–2016), a cognitive battery was administered by trained interviewers and included the Digit Symbol Substitution Test (DSST) that assesses processing speed, executive function, and working memory²⁷; the Stroop test that measures executive function²⁸; the Rey Auditory Verbal Learning Test (RAVLT) that measures delayed verbal memory²⁹; the Montreal Cognitive Assessment (MoCA) that assesses global cognitive function³⁰; and Letter Fluency and Category Fluency tests that were combined to assess fluency. Higher scores indicate a better function for all tests, except for the Stroop test, for which higher scores indicate worse performance. We defined poor cognitive performance as a score that was >1 SD below the mean score.

Covariates

Potential confounders were determined at baseline (for the purposes of this study, the baseline was defined as the Year 15 follow-up examination, which took place between 2000 and 2001). Age, sex, race, educational attainment, and smoking were assessed using self-administered and interviewer-administered questionnaires. Physical activity score was derived from the self-reported frequency and intensity of 13 activities in the previous year in exercise units.³¹ Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared. Self-reported depressive symptoms in the past week were examined using the Center for Epidemiologic Studies Depression scale.³² We defined diabetes diagnosis as fasting glucose level ≥126 mg/dL or taking antidiabetic medication, and hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or taking antihypertensive medication.

Statistical Analysis

Baseline characteristics were compared by the categories of sleep duration (short: <6 hours; average: 6–7 hours; long: >7 hours), tertiles of sleep fragmentation, and according to subjective sleep quality (good vs poor), using Student t tests, 1-way ANOVA, Mann-Whitney U test, or Kruskal-Wallis test for continuous variables and chi-square tests for categorical variables. We used multivariable logistic regression to examine the association between the categories of sleep duration, sleep fragmentation index, or subjective sleep quality and poor cognitive performance in midlife, adjusting for age, sex, race, and education, and then with additional adjustment for smoking, BMI, depression, physical activity, hypertension, and diabetes. In secondary analysis, we explored the interaction by race and sex. We also additionally adjusted for the risk of sleep apnea to explore whether the association was due to the presence of sleep apnea. The results are presented as odds ratio (OR) with 95% confidence intervals. All statistical tests were two-sided, and a p value of <0.05 was considered significant. Analyses were performed using Stata, version 14.1 (Stata Corp LP, College Station, TX), and in the R 4.2.2 statistical computing environment.

Data Availability

Anonymized data can be accessed from the CARDIA Coordinating Center (cardia.dopm.uab.edu/contact-cardia). A description of the National Heart, Lung, and Blood Institute policies governing the data and describing access to the data can be found online (cardia.dopm.uab.edu/study-information/nhlbi-data-repository-data).

Results

Our study sample included 305 (58%) women and 229 (44%) Black adults, with a mean age of 40.1 ± 3.5 years at baseline. On average, participants slept 6.1 ± 1.1 hours and had a fragmentation index of 19.2 ± 8.1%, with 239 (45.6%) reporting poor sleep as defined by a PSQI global score of >5. Table 1 summarizes baseline characteristics according to the sleep parameters. Those who had actigraphy-measured shorter sleep duration or higher sleep fragmentation were significantly more likely to be male, Black adults, had higher BMI, and were more likely to have a history of depression or hypertension. In total, 142 (64.3%) participants with shorter (<6 hours) sleep duration vs 17 (17.5%) of those with long (>7 hours) sleep were Black adults. Those who had poorer subjective sleep quality were more likely to be younger, female, Black adults, had lower education, higher BMI, lower physical activity, and were more likely to have a history of depression or hypertension.

Table 1.

Baseline Characteristics of the Study Participants by Objective Sleep Duration and Quality

	Total	Sleep duration				Sleep fragmentation
			Short (<6 h)	Average (6–7 h)	Long (>7 h)		Low (<15.3)	Moderate (15.3–20.6)	High (>20.6)
		p Value	N = 176	N = 176	N = 174	p Value	N = 176	N = 175	N = 175
Age, y	40.1 ± 3.5	0.62	40.0 ± 3.7	40.3 ± 3.5	39.9 ± 3.4	0.56	40.5 ± 3.5	40.2 ± 3.5	39.6 ± 3.7
Female	305 (58.0)	<0.001	109 (49.3)	120 (57.7)	76 (78.4)	<0.001	121 (68.8)	104 (59.4)	80 (45.7)
Black race	229 (43.5)	<0.001	142 (64.3)	70 (33.7)	17 (17.5)	<0.001	58 (33.0)	71 (40.6)	100 (57.1)
High school and higher	421 (80.0)	0.1	162 (73.3)	179 (86.1)	80 (82.5)	<0.001	156 (88.6)	141 (80.6)	124 (70.9)
Body mass index	28.3 ± 6.8	<0.001	29.9 ± 7.5	27.7 ± 5.8	26.0 ± 6.5	<0.01	27.1 ± 6.5	28.6 ± 6.4	29.3 ± 7.4
Depression	85 (16.4)	0.02	48 (22.0)	26 (12.6)	12 (12.4)	0.03	19 (10.9)	30 (17.2)	37 (21.4)
Higher physical activity	263 (50.0)	0.02	95 (43.0)	117 (56.3)	51 (52.6)	0.30	95 (54.0)	88 (50.3)	80 (45.7)
Hypertension	54 (10.3)	0.06	30 (13.6)	19 (9.1)	5 (5.2)	0.04	11 (6.3)	18 (10.3)	25 (14.3)
Diabetes	11 (2.1)	0.43	7 (3.2)	3 (1.5)	1 (1.0)	0.08	3 (1.7)	1 (0.6)	7 (4.1)

Results are presented as mean ± SD or N (%).

Table 2 presents the OR (95% CI) of cognitive impairment by sleep parameters after adjustment for age, sex, race, and education. Compared with participants in the lowest tertile of sleep fragmentation index (<15.3), those in the highest tertile were 3 times as likely to have poor cognitive performance at midlife on the DSST (OR = 3.32; 95% CI 1.52–7.21) and had more than double the odds of poor cognitive performance on the Stroop (OR = 2.29; 95% CI 1.09–4.83), fluency (OR = 2.78; 95% CI 1.37–5.64), and MOCA test (OR = 2.20; 95% CI 1.05–4.59), but not on RAVLT (OR = 1.39; 95% CI 0.76–2.54). There were no differences in cognitive performance at midlife for those in moderate sleep fragmentation index categories compared with the lowest tertile. Compared with participants who reported good sleep quality (PSQI ≤5), those who reported poor sleep quality (PSQI >5) were 83% more likely to have poor cognitive performance on the Stroop test (OR = 1.83; 95% CI 1.02–3.26). There was no association between actigraphy-measured sleep duration and any measure of cognitive performance.

Table 2.

Odds Ratios^a (95% CI) of Poor Cognitive Performance by Sleep Parameters in Midlife

	DSST	Stroop	Fluency	MOCA	RAVLT
N (%) with poor cognitive performance	78 (14.8%)	69 (13.1%)	80 (15.2%)	83 (15.8%)	108 (20.5%)
Sleep duration
Average (6–7 h)	Reference
Short (<6 h)	0.85 [0.47, 1.55]	1.89 [0.96, 3.71]	1.06 [0.63, 1.77]	0.90 [0.49, 1.65]	0.88 [0.49, 1.56]
Long (>7 h)	0.88 [0.36, 2.18]	2.43 [0.92, 6.43]	0.96 [0.44, 2.07]	0.78 [0.29, 2.09]	0.86 [0.38, 1.94]
Sleep fragmentation
Low (<15.3)	Reference
Moderate (15.3–20.6)	2.07 [0.92, 4.64]	1.17 [0.52, 2.6]	1.81 [0.88, 3.72]	1.72 [0.81, 3.68]	1.41 [0.77, 2.57]
High (>20.6)	3.32 [1.52, 7.21]c	2.29 [1.09, 4.83]b	2.78 [1.37, 5.64]c	2.2 [1.05, 4.59]b	1.39 [0.76, 2.54]
Subjective sleep quality
Good (PSQI ≤5)	Reference
Poor (PSQI >5)	1.59 [0.92, 2.73]	1.83 [1.02, 3.26]b	1.68 [1.00, 2.84]	1.56 [0.89, 2.71]	1.19 [0.74, 1.89]

Abbreviations: DSST = Digit Symbol Substitution Test; MOCA = Montreal Cognitive Assessment; PSQI = Pittsburgh Sleep Quality Index; RAVLT = Rey Auditory Verbal Learning Test.

^aAdjusted for age, sex, race, education.

^bp < 0.05.

^cp < 0.01.

After further adjustment for BMI, depression, physical activity, hypertension, and diabetes at baseline, actigraphy-measured higher (vs lower) sleep fragmentation remained significantly associated with poor cognitive performance on the DSST (OR = 2.97; 95% CI 1.34–6.56), fluency (OR = 2.42; 95% CI 1.17–5.02), and MOCA test (OR = 2.29; 95% CI 1.06–4.94), while the association for the Stroop test was slightly attenuated (OR = 2.10; 95% CI 0.97–4.58). Sleep fragmentation was not associated with poor cognitive performance on the RAVLT (highest vs lowest tertile: OR = 1.23; 95% CI 0.66–2.29). Again, there was no association between moderate compared with lower sleep fragmentation and cognitive performance. The association between sleep fragmentation and cognitive performance did not differ by race or sex. Adjustment for the risk of sleep apnea did not appreciably alter the results. No association was observed between actigraphy-measured sleep duration, subjective sleep quality, or subjective sleep duration assessed by PSQI and cognitive performance. Table 3 presents the multivariable adjusted ORs (95% CI) of poor cognitive performance on each test by sleep parameters. eTable 1 (links.lww.com/WNL/D293) presents the adjusted mean scores of cognitive function by sleep parameters.

Table 3.

Fully Adjusted Odds Ratios^a (95% CI) of Poor Cognitive Performance by Sleep Parameters in Midlife

	DSST	Stroop	Fluency	MOCA	RAVLT
N (%) with poor cognitive performance	78 (14.8%)	69 (13.1%)	80 (15.2%)	83 (15.8%)	108 (20.5%)
Sleep duration
Average (6–7 h)	Reference
Short (<6 h)	0.77 [0.41, 1.42]	1.62 [0.81, 3.25]	0.76 [0.42, 1.38]	0.92 [0.49, 1.74]	0.93 [0.54, 1.58]
Long (>7 h)	0.91 [0.36, 2.28]	2.60 [0.97, 6.99]	0.83 [0.36, 1.90]	0.81 [0.30, 2.20]	0.94 [0.43, 2.05]
Sleep fragmentation
Low (<15.3)	Reference
Moderate (15.3–20.6)	1.93 [0.85, 4.38]	1.03 [0.45, 2.35]	1.81 [0.87, 3.76]	1.68 [0.76, 3.70]	1.34 [0.72, 2.47]
High (>20.6)	2.97 [1.34, 6.56]c	2.10 [0.97, 4.58]	2.42 [1.17, 5.02]b	2.29 [1.06, 4.94]b	1.23 [0.66, 2.29]
Subjective sleep quality
Good (PSQI ≤5)	Reference
Poor (PSQI>5)	1.45 [0.81, 2.63]	1.70 [0.90, 3.21]	1.55 [0.87, 2.75]	1.31 [0.71, 2.41]	1.07 [0.64, 1.80]

Abbreviations: DSST = Digit Symbol Substitution Test; MOCA = Montreal Cognitive Assessment; PSQI = Pittsburgh Sleep Quality Index; RAVLT = Rey Auditory Verbal Learning Test.

^aAdjusted for age, sex, race, education, BMI, depression, physical activity, hypertension, diabetes.

^bp < 0.05.

^cp < 0.01.

Discussion

Among middle-aged Black and White adults, actigraphy-measured high sleep fragmentation in early midlife was independently associated with worse executive function, fluency, and global cognition approximately 11 years later; a threshold seems to exist with the degree of sleep fragmentation. No association was found for sleep duration or self-reported sleep quality. Our findings suggest that the association between sleep quality and cognition may become prominent as early as in midlife. Furthermore, this study highlights the importance of using objective measures of sleep and suggests that it is the quality rather than quantity of sleep that is particularly important for cognitive health in midlife.

Over the past 2 decades, there has been a surge of epidemiologic studies and meta-analyses that confirm a robust association between sleep disturbances, such as sleep apnea and disrupted circadian rhythms, and increased risk of cognitive impairment in older adults.^4-6,33-36 However, the association of midlife sleep characteristics and cognition is understudied and controversial. One recent study found that short sleep duration in late midlife as assessed by both questionnaires and accelerometer was associated with incident dementia over 25 years.¹⁷ A few studies focused on self-reported sleep characteristics in midlife and found that subjective short and long sleep duration, poor sleep quality, or excessive daytime sleepiness in midlife was associated with poorer cognitive performance in midlife or late life.^14,15 Findings from 4 population-based studies in Sweden and Finland reported that self-reported sleep problems in late life rather than in midlife were associated with poorer global cognition.¹⁶ Notably, subjective measures of sleep quality are susceptible to measurement errors and might not have the same implications as objective measures.^18,19,37

We examined both sleep duration and quality, measured both subjectively and objectively by actigraphy, in early midlife and their relation to performance on several cognitive domains in late midlife. We found that actigraphy-measured greater sleep fragmentation was independently associated with worse cognition in midlife, whereas there was no association between sleep duration or self-reported sleep quality and cognition. These results are in line with a few previous studies suggesting a significant relationship between cognition and objective sleep quality rather than sleep duration in older adults.^20,38 However, several studies showed a U-shaped relationship between self-reported sleep duration and cognition, with both short and long sleep duration associated with poorer cognition performance and greater risk of dementia.^14,15,39 It is possible that sleep duration and particularly self-reported extreme sleep duration might be a symptom of age-related cognitive impairment, and thus, the association between sleep duration and cognitive impairment might be more pronounced in older adults. Indeed, findings from our previous Mendelian randomization study suggested that genetic liability for AD predicted short sleep duration only among those aged 55 years and older,⁴⁰ indicating the role of short sleep duration as an early marker of AD in older adults. Sleep duration in midlife may be influenced by a variety of factors, such as age, life choice, or work stress, and thus may have a more complex relationship with cognitive performance. Prior work from the CARDIA study also suggested that self-reported short sleep in midlife, although associated with white matter abnormality on brain MRI, may not yet be sufficient to affect cognition.⁴¹ Furthermore, the average sleep duration identified in this study was shorter than the duration reported by previous studies that used subjective measures and that found a significant association between sleep duration and cognition.¹⁴ Indeed, our earlier work from the CARDIA study showed that self-reported measures overestimated sleep duration by 0.8 hours compared with actigraphy measures.¹⁸ This could also have led to discrepancies in findings related to sleep duration.

While the mechanisms through which sleep is linked to AD have been extensively studied¹ and may involve the activation of the glymphatic system with subsequent clearance of brain interstitial metabolic waste products during sleep,⁴² less is known about the pathways linking sleep fragmentation to cognition in midlife. Experimental and epidemiologic studies suggested that sleep deprivation or self-reported poor sleep quality led to higher CSF amyloid β (Aβ) levels in middle-aged adults.^43-45 Interestingly, we found a robust association between sleep fragmentation and executive function or fluency but not verbal memory. This adds to previous studies of older adults suggesting a strong association between executive function and sleep disorders, including sleep-disordered breathing (SDB),^6,46 which might be due to the influence of sleep on the prefrontal cortex area. In addition, sleep fragmentation in midlife may be particularly relevant to a greater risk of cardiometabolic diseases.^47-49 This is further supported by the accumulating evidence suggesting the impact of cardiovascular risk factors on accelerated brain aging in midlife, especially in subcortical frontal regions.⁵⁰

This study has several strengths, including the assessment of both sleep and cognition in midlife, the use of both self-reported and actigraph measures of sleep duration and quality, the examination of a diverse sample including both Black and White adults, and the examination of multiple cognitive domains. However, there are also several limitations to consider. One limitation is the lack of polysomnography measures of sleep, the gold standard measure that would provide more detailed and accurate information on sleep quality and certain sleep disorders, particularly SDB. We have collected actigraphy data for 3 days on 2 occasions, approximately 1 year apart, to obtain 6 days of sleep data to maximize feasibility and sample size. Given the limited number of days for data collection and the variability in work schedules, we could not adjust for weekday or weekend differences. There may also be residual confounding from other factors that could affect both sleep and cognition. Given the relatively small sample size, we have limited statistical power to fully investigate potential race or sex differences. Although we examined sleep in the mid-30s to late 40s and cognition 11 years later, we were unable to control for cognition at baseline or study the link between sleep and cognition beyond this period of life. We also could not assess whether cognition improved with an improvement in sleep quality. Future studies are needed to determine the impact of sleep on cognition even earlier in life and clarify the direction of this relationship.

In conclusion, we found that actigraphy-measured sleep fragmentation in early midlife was associated with late midlife cognition among Black and White adults. Our findings indicate that the quality rather than the quantity of sleep matters most for cognitive health in midlife and that the measures of sleep should go beyond self-report. Although this study reported on sleep and cognition in a middle-aged diverse population, larger studies are required to determine if race or sex differences exist. Importantly, this research contributes to an exciting body of literature that stresses the need for a life course approach to evaluate modifiable risk factors for cognitive aging. While evidence is emerging for a link between midlife sleep and late life cognitive outcomes, our findings highlight the importance of studying both sleep and cognition earlier in life. Given the long preclinical window of AD and the high prevalence of sleep problems, the understanding of midlife sleep disturbances either as an early behavioral marker or modifiable risk factor for ADRD has significant public health implications. Future research is needed to assess the association between sleep disturbances and cognition at different stages of life and to identify whether critical life periods exist when sleep is more strongly associated with cognition. Additional intervention studies that examine the impact of sleep treatment on cognition at different periods of life might open up new opportunities for the prevention of AD in late life.

Glossary

AD

Alzheimer disease

BMI

body mass index

CARDIA

Coronary Artery Risk Development in Young Adults

DSST

Digit Symbol Substitution Test

MoCA

Montreal Cognitive Assessment

OR

odds ratio

PSQI

Pittsburgh Sleep Quality Index

RAVLT

Rey Auditory Verbal Learning Test

SDB

sleep-disordered breathing

Appendix. Authors

Name	Location	Contribution
Yue Leng, PhD	Department of Psychiatry and Behavioral Sciences, University of California, San Francisco	Drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data
Kristen Knutson, PhD	Department of Neurology, Northwestern University Feinberg School of Medicine, Chicago, IL	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Mercedes R. Carnethon, PhD	Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data
Kristine Yaffe, MD	Departments of Psychiatry and Behavioral Sciences, Neurology, and Epidemiology, University of California, San Francisco; VA Medical Center, San Francisco, CA	Drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design

Study Funding

The Coronary Artery Risk Development in Young Adults Study (CARDIA) is supported by contracts HHSN268201800003I, HHSN268201800004I, HHSN268201800005I, HHSN268201800006I, and HHSN268201800007I from the National Heart, Lung, and Blood Institute (NHLBI). Y. Leng is supported by NIA R00 AG056598. The CARDIA Cognitive Function ancillary study is supported by NHLBI grant R01 HL122658. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.