Abstract
This study aimed to explore the dose-response relationship between sleep duration and phenotypic age acceleration (PhenoAgeAccel), as well as the potential impact of different sleep duration patterns on the biological aging process. Utilizing data from the National Health and Nutrition Examination Survey from 2001 to 2010, this cross-sectional study included 8992 adult participants. Sleep duration data were collected via self-report and categorized into “<7 hours/day,” “7–9 hours/day,” and “>9 hours/day” groups. PhenoAgeAccel was calculated by combining actual age with 9 biomarkers. Weighted generalized linear regression models and unrestricted cubic spline analyses were employed to examine the relationship between sleep duration and PhenoAgeAccel. Interaction effects were assessed to evaluate the influence of different demographic and health characteristics. In unadjusted analyses, the 7 to 9 hours/day sleep group showed a significant deceleration in phenotypic aging compared to the <7 hours/day group (β = −1.207, P < .0001). However, this association was substantially attenuated and no longer statistically significant after full adjustment for demographic, lifestyle, and comorbidity factors. A significant nonlinear dose-response relationship was confirmed, with an inflection point at approximately 6.7 hours. Interaction effect tests revealed that this relationship was significantly influenced by an individual’s smoking and diabetes status (P < .01). This study suggests that moderate sleep duration of 7 to 9 hours/day is associated with a deceleration in phenotypic aging, with a critical inflection point for sleep and aging health at approximately 6.7 hours. Both insufficient and excessive sleep durations may be detrimental to slowing the aging process. The results of the interaction effect tests emphasize the need to consider individual smoking and diabetes status when developing targeted health interventions. These findings provide new insights into the complex relationship between sleep and biological aging, offering a scientific basis for public health guidance and the optimization of individual sleep habits. Future research should employ longitudinal designs and objective sleep monitoring tools to further explore the causal relationship and underlying mechanisms between sleep and biological aging.
Keywords: aging, phenotypic age acceleration, sleep duration
1. Introduction
In contemporary society, sleep issues have emerged as a global public health challenge, particularly concerning their impact on the aging process. It is well-established that sufficient and high-quality sleep is crucial for maintaining health and preventing premature aging. However, with the acceleration of life’s pace and increased societal pressures, an increasing number of individuals are experiencing the plight of insufficient sleep. Studies have shown that lack of sleep is closely associated with a variety of health problems, such as cardiovascular diseases, diabetes, obesity, and depression, all of which may accelerate the aging process of individuals.[1]
Aging is a complex biological process involving changes in various physiological and molecular mechanisms. In recent years, phenotypic age acceleration (PhenoAgeAccel), as an important indicator for measuring the speed of an individual’s biological aging, has garnered widespread attention in the scientific community. PhenoAgeAccel predicts the speed of an individual’s biological aging by assessing age-related biomarkers, providing a new perspective for understanding individual health conditions. For instance, research by Chen et al[2] demonstrated that PhenoAgeAccel is positively associated with all-cause and specific-cause mortality in diabetes patients, highlighting the potential value of antiaging treatments in slowing disease progression; the study by Shadyab et al[3] emphasized the potential value of PhenoAgeAccel as a biomarker in assessing cognitive impairment and brain structure, as well as whether these associations are influenced by cardiovascular diseases.
The relationship between sleep and aging is particularly complex.[4–7] On one hand, changes in sleep quality and duration directly affect an individual’s physiological and psychological health, which may in turn accelerate the aging process. On the other hand, aging itself may affect sleep patterns, leading to the occurrence of sleep problems. Therefore, understanding the relationship between sleep and PhenoAgeAccel is crucial for developing effective public health strategies and interventions.
After an in-depth analysis of the key role of sleep duration in the biological aging process, we further explored the connection between sleep patterns and PhenoAgeAccel, as well as the importance of these factors in disease risk assessment and aging research. We recognize the significant potential of these indicators in revealing the subtle mechanisms of aging and in formulating effective health intervention strategies. However, despite some key findings, the dose-response relationship between sleep patterns and PhenoAgeAccel has not been fully clarified.
Thus, this study aims to delve into the relationship between sleep duration and PhenoAgeAccel and attempts to reveal the potential synergistic mechanisms between sleep patterns and the biological aging process. Our preliminary hypothesis is that there is a significant correlation between sleep duration and PhenoAgeAccel, suggesting that optimizing sleep patterns may contribute to slowing the process of biological aging.
2. Materials and methods
2.1. Study subjects and data sources
This investigation utilized data from the National Health and Nutrition Examination Survey (NHANES) conducted between 2001 and 2010. Orchestrated by the Centers for Disease Control and Prevention (CDC), NHANES is a representative national survey designed to gauge the health and nutritional status of American adults and children. The cohort of participants was selected from the NHANES database, in alignment with the protocols recommended by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Administered every 2 years by the National Center for Health Statistics alongside the Centers for Disease Control and Prevention, the survey employs a complex, multi-stage probability sampling method to examine around 10,000 noninstitutionalized individuals from various US locales. Data collection involved household interviews and physical examinations, where participants answered questions related to demographic, socioeconomic, dietary, and health-related factors, and underwent assessments for medical, dental, and physiological-biochemical indicators.
In this study, a total of 8992 adults aged ≥20 were included in the final analysis after a rigorous screening process. This process primarily involved the exclusion of participants who did not meet the age criteria and those who failed to complete the requisite questionnaires and physical examinations. Data analysis was conducted using a complete case analysis approach. The detailed screening procedure is depicted in Figure 1. The NHANES survey data utilized in this study has undergone the necessary ethical approval, adhering to the principles of the Declaration of Helsinki.
Figure 1.
Flowchart. This figure illustrates the detailed process of participant inclusion and exclusion criteria based on NHANES data. Participants who did not meet the age criteria and those who failed to complete the requisite questionnaires and physical examinations were excluded, leading to a final analytical sample of 8992 adults. NHANES = National Health and Nutrition Examination Survey.
2.2. Definition of the sleep time
In the NHANES dataset, the measurement of sleep duration was obtained through a questionnaire survey. Participants were asked, “How much do you usually sleep during a typical 24-hour period on weekdays or workdays?” This question aimed to capture their typical sleep habits. The response options ranged from 1 hour to 12 hours, with 12 hours representing a sleep duration of 12 hours or more. This self-reported sleep duration data were used to analyze the relationship between sleep and various health outcomes. In this study, sleep duration was categorized into “<7 hours/day,” “7–9 hours/day,” and “>9 hours/day” groups. This categorization helps in more accurately assessing the relationship between different durations of sleep and PhenoAgeAccel.
2.3. Definition of the PhenoAgeAccel
This study adopted a comprehensive approach to calculate an individual’s phenotypic age, based on pivotal research on phenotypic age.[8] This method considers both chronological age and 9 key biomarkers, including albumin, creatinine, glucose, log-transformed C-reactive protein, percentage of lymphocytes, mean cell volume, red cell distribution width, alkaline phosphatase, and white blood cell count. The selection of these biomarkers was based on a Cox proportional hazards elastic net model for mortality, validated through 10-fold cross-validation. The study employed 2 Gompertz proportional hazards models to parameterize phenotypic age: one model incorporated all 10 selected variables, while the other considered only chronological age. This calculation method aids in a more precise assessment of an individual’s biological aging process.
Particular attention was given to PhenoAgeAccel, an indicator derived from the residuals of a linear regression analysis comparing phenotypic age against chronological age.[9–13] For instance, if 2 individuals of the same age display different levels of physiological health and vitality, leading one to appear younger and the other older due to health issues or poor lifestyle choices, their PhenoAgeAccel values would differ. As a lagging indicator, lower PhenoAgeAccel values signify a slower process of biological aging, offering a crucial perspective on the variance of an individual’s physiological state relative to their chronological age. The specific formula for PhenoAgeAccel is as follows:
where xb = −19.907 − 0.0336 × albumin + 0.0095 × creatinine + 0.0195 × glucose + 0.0954 × ln(CRP) − 0.0120 × lymphocyte percent + 0.0268 × mean cell volume + 0.3356 × red cell distribution width + 0.00188 × alkaline phosphatase + 0.0554 × white blood cell count + 0.0804 × chronological age.
2.4. Covariate
This investigation incorporated a variety of covariates for a comprehensive analysis, encompassing fundamental demographic information, lifestyle factors, and health status indicators. Demographic data included age, gender (male, female), race (Mexican American, non-Hispanic Black, non-Hispanic White, other), and education levels (less than high school, high school, beyond high school). Economic status was gauged through the poverty income ratio (PIR), comparing household or individual income against the poverty guidelines for the survey year (low income PIR ≤ 1.3, middle income 1.3 < PIR < 3.5, high income ≥ 3.5).
Lifestyle factors covered marital status (married/living with partner, single, widowed/divorced/separated), body mass index (BMI) classifications (<25, 25–29.9, ≥30 kg/m2), smoking status (former smoker, nonsmoker, current smoker), and alcohol consumption habits (never, former, light, moderate, heavy). Physical activity was assessed through the typical time and energy expenditure in physical activities over the past week, including vigorous and moderate activities during work, commute, and leisure, quantified using metabolic equivalent of task (METs h/week) scores. Dietary quality was evaluated with the Healthy Eating Index-2020 (HEI-2020), measuring the alignment of an individual or group’s diet with the Dietary Guidelines for Americans issued by the US Department of Agriculture. The HEI-2020, updated to reflect current nutritional recommendations and dietary patterns, encompasses several components representing vital aspects of the diet such as fruits, vegetables, whole grains, dairy, proteins, the ratio of unsaturated to saturated fats, and intake of sodium and added sugars. The scoring system of HEI-2020 is quantitative, with a total score of 100 points, where higher scores indicate better dietary quality and greater adherence to the Dietary Guidelines for Americans.[14]
Health status indicators included diagnostic criteria for hypertension, hyperlipidemia, and diabetes. Hypertension was determined based on physician diagnosis, antihypertensive medication usage, and blood pressure measurements (systolic ≥140 mm Hg or diastolic ≥90mm Hg). Hyperlipidemia diagnosis relied on lipid levels (triglycerides, total cholesterol, low-density lipoprotein, high-density lipoprotein) and the use of lipid-lowering medications. Diabetes assessment was based on physician diagnosis, levels of glycated hemoglobin, fasting glucose, and the use of antidiabetic medications. Cancer prevalence was determined by diagnoses from physicians or health professionals.
2.5. Statistical methods
This study meticulously adhered to the complex survey sampling procedures of the NHANES and computed complex sampling weights as per the analysis guidelines provided by NHANES. To ensure the national representativeness of the results, all statistical analyses were conducted using weighted data. Continuous variables were reported as means and their standard errors (mean ± SE), while categorical variables were presented as counts and their weighted percentages.
To assess the differences between groups, the study utilized 1-way analysis of variance for continuous variables and chi-square tests for categorical variables. Additionally, weighted generalized linear regression models were applied to examine the linear relationship between sleep duration and PhenoAgeAccel, with subgroup analyses conducted for gender to investigate the impact of sex differences on this relationship.
Furthermore, based on the outcomes of the linear regression models, this study employed restricted cubic splines to test for nonlinear trends between variables. To further validate the influence of control variables on the relationship between sleep duration and PhenoAgeAccel, these variables were incorporated into the interaction effect test models.
All statistical analyses were conducted with a 2-tailed P-value <.05 considered statistically significant, and data processing and analysis were performed using R Studio (version 4.2.1, USA).
3. Result
3.1. Baseline characteristics of the study population
Table 1 displays characteristics of the study population grouped by sleep duration within the 2001 to 2010 NHANES dataset, revealing significant differences in sleep duration and PhenoAgeAccel across different groups (P < .01). Significant variances were also observed in the distribution of gender, age, race, BMI, marital status, education level, PIR, smoking status, alcohol consumption, physical activity, dietary quality, diabetes, hypertension, and cancer across various sleep duration categories (P < .0001). However, the distribution of hyperlipidemia did not show significant differences across the sleep groups (P = .84).
Table 1.
Baseline characteristics of the study population.
| Characteristic | Overall | <7 hours/day | 7–9 hours/day | >9 hours/day | P-value |
|---|---|---|---|---|---|
| N | 8992 | 3515 | 4915 | 562 | |
| PhenoAgeAccel | −4.86 (0.08) | −3.57 (0.14) | −4.51 (0.13) | −5.08 (0.12) | <.0001 |
| Gender, n (weighted %) | <.001 | ||||
| Female | 4179 (48.28) | 1576 (45.63) | 2318 (49.02) | 285 (57.08) | |
| Male | 4813 (51.72) | 1939 (54.37) | 2597 (50.98) | 277 (42.92) | |
| Age (yr), n (weighted %) | <.0001 | ||||
| 20–29 | 1493 (17.87) | 542 (17.02) | 827 (17.84) | 124 (23.33) | |
| 30–39 | 1675 (19.48) | 690 (20.60) | 909 (19.28) | 76 (14.62) | |
| 40–49 | 1706 (22.21) | 729 (24.11) | 888 (21.14) | 89 (21.28) | |
| 50–59 | 1464 (19.98) | 627 (20.82) | 779 (20.27) | 58 (11.94) | |
| ≥60 | 2654 (20.46) | 927 (17.45) | 1512 (21.47) | 215 (28.83) | |
| Race, n (weighted %) | <.0001 | ||||
| Non-Hispanic White | 4846 (75.15) | 1602 (67.98) | 2911 (79.44) | 333 (76.18) | |
| Mexican American | 1488 (6.87) | 545 (6.81) | 865 (7.01) | 78 (5.92) | |
| Non-Hispanic Black | 1606 (9.14) | 884 (14.18) | 623 (5.93) | 99 (10.27) | |
| Other race (including multi-racial and other Hispanic) | 1052 (8.84) | 484 (11.03) | 516 (7.62) | 52 (7.63) | |
| Body mass index (kg/m2), n (weighted %) | <.0001 | ||||
| <25 | 2683 (32.38) | 936 (29.20) | 1545 (33.52) | 202 (40.38) | |
| 25–29.9 | 3092 (33.71) | 1178 (32.51) | 1741 (34.78) | 173 (30.38) | |
| ≥30 | 3217 (33.91) | 1401 (38.29) | 1629 (31.70) | 187 (29.23) | |
| Marital status, n (weighted %) | <.0001 | ||||
| Married/living with partner | 5713 (67.53) | 2117 (64.34) | 3266 (70.16) | 330 (60.93) | |
| Never married | 1460 (15.82) | 600 (16.57) | 739 (14.72) | 121 (22.25) | |
| Widowed/divorced/separated | 1819 (16.64) | 798 (19.09) | 910 (15.12) | 111 (16.82) | |
| Education level, n (weighted %) | <.0001 | ||||
| Below | 2081 (14.79) | 842 (16.26) | 1090 (13.47) | 149 (18.79) | |
| High school | 2114 (23.24) | 908 (26.59) | 1071 (20.98) | 135 (25.28) | |
| Above | 4797 (61.97) | 1765 (57.15) | 2754 (65.55) | 278 (55.93) | |
| Poverty to income ratio, n (weighted %) | <.0001 | ||||
| <1.3 | 2385 (16.57) | 1025 (19.69) | 1200 (14.23) | 160 (20.78) | |
| 1.3–3.49 | 3338 (33.90) | 1319 (35.03) | 1785 (32.75) | 234 (38.46) | |
| ≥3.5 | 3269 (49.53) | 1171 (45.28) | 1930 (53.02) | 168 (40.76) | |
| Smoke status, n (weighted %) | <.0001 | ||||
| Former smoker | 2313 (25.44) | 825 (23.65) | 1341 (26.81) | 147 (22.80) | |
| Nonsmoker | 4648 (52.38) | 1730 (48.46) | 2646 (54.99) | 272 (50.30) | |
| Current smoker | 2031 (22.18) | 960 (27.88) | 928 (18.20) | 143 (26.90) | |
| Alcohol status, n (weighted %) | <.0001 | ||||
| Former | 1573 (14.54) | 701 (16.74) | 772 (13.12) | 100 (15.26) | |
| Never | 992 (8.96) | 372 (9.06) | 541 (8.51) | 79 (12.82) | |
| Mild | 3021 (36.85) | 1103 (34.43) | 1740 (38.89) | 178 (31.47) | |
| Moderate | 1424 (17.44) | 541 (16.17) | 799 (18.08) | 84 (18.83) | |
| Heavy | 1982 (22.20) | 798 (23.60) | 1063 (21.40) | 121 (21.62) | |
| Physical activity (MET-min/wk), n (weighted %) | <.0001 | ||||
| Q1 [0.93, 700] | 3005 (34.30) | 1124 (32.20) | 1677 (35.16) | 204 (38.58) | |
| Q2 [700, 2880] | 3070 (34.97) | 1138 (32.61) | 1724 (36.21) | 208 (36.92) | |
| Q3 [2880, 55,440] | 2917 (30.73) | 1253 (35.18) | 1514 (28.62) | 150 (24.50) | |
| Healthy eating index, n (weighted %) | <.0001 | ||||
| Q1 [0, 44.31] | 2998 (33.96) | 1312 (38.94) | 1495 (30.88) | 191 (34.14) | |
| Q2 [44.31, 56.49] | 2996 (33.53) | 1180 (33.93) | 1641 (33.63) | 175 (30.09) | |
| Q3 [56.49, 96.11] | 2998 (32.51) | 1023 (27.13) | 1779 (35.49) | 196 (35.78) | |
| Diabetes, n (weighted %) | .002 | ||||
| No | 6876 (81.10) | 2633 (79.00) | 3828 (82.56) | 415 (79.48) | |
| Yes | 2116 (18.90) | 882 (21.00) | 1087 (17.44) | 147 (20.52) | |
| Hypertension, n (weighted %) | .001 | ||||
| No | 5512 (65.86) | 2052 (63.07) | 3112 (67.65) | 348 (65.06) | |
| Yes | 3480 (34.14) | 1463 (36.93) | 1803 (32.35) | 214 (34.94) | |
| Hyperlipidemia, n (weighted %) | .84 | ||||
| No | 2461 (28.11) | 973 (27.64) | 1331 (28.35) | 157 (28.64) | |
| Yes | 6531 (71.89) | 2542 (72.36) | 3584 (71.65) | 405 (71.36) | |
| Cancer, n (weighted %) | .01 | ||||
| No | 8189 (91.34) | 3237 (92.31) | 4458 (91.04) | 494 (88.47) | |
| Yes | 803 (8.66) | 278 (7.69) | 457 (8.96) | 68 (11.53) |
The continuity variables involved in this study are expressed by means (standard error), and the categorical variables are expressed by actual quantities (weighted percentages). The 1-way ANOVA applies to continuity variables and Chi-square test applies to categorical variables.
ANOVA = analysis of variance, MET = metabolic equivalent of task, Q1 = the first quartile, Q2 = the second quartile, Q3 = the third quartile.
Notably, the PhenoAgeAccel overlapping density plot illustrated in Figure 2. indicates that the distribution peak for the 7 to 9 hours/day sleep group is closer to the 0 point, suggesting that this group’s PhenoAgeAccel is closer to the average level (with less variability), implying a higher likelihood of actual age matching biological age. In contrast, the curves for the <7 hours/day and >9 hours/day sleep groups deviate further from the 0 point, indicating that the daily 7 to 9 hours sleep duration group tends to have a more concentrated distribution with less fluctuation in PhenoAgeAccel. This may imply that this range of sleep duration is more conducive to maintaining a health status that corresponds with actual age, thereby decelerating the biological aging process.
Figure 2.
Distribution of phenotypic age acceleration by sleep group. This figure shows the density distribution of phenotypic age acceleration for the 3 sleep duration groups (<7 hours/day, 7–9 hours/day, and >9 hours/day). The x-axis represents the PhenoAgeAccel value, and the y-axis represents the density of the participant distribution.
3.2. Association analysis between sleep time and PhenoAgeAccel in American adults
This study investigated the relationship between daily sleep duration and PhenoAgeAccel, with the results depicted in Figure 3. In the unadjusted Crude Model, the 7 to 9 hours/day sleep group exhibited a significant deceleration in phenotypic aging compared to the reference group with <7 hours of sleep (β = −1.207; 95% CI: −1.599, −0.815; P < .0001). After adjusting for key demographic and lifestyle factors in Model 1, this association remained statistically significant (β = −0.373; 95% CI: −0.737, −0.009; P = .045). However, in the fully adjusted Model 2, the protective trend for the 7 to 9 hour group was attenuated and no longer statistically significant (β = −0.251; P = .166). Meanwhile, the association between sleeping >9 hours/day and accelerated phenotypic aging became evident after adjustments, reaching statistical significance in both Model 1 (P = .006) and Model 2 (P = .009).
Figure 3.
Association analysis between sleep duration and PhenoAgeAccel in American adults. This figure displays the association between sleep duration and PhenoAgeAccel. The forest plot (left) shows the regression coefficients (β) and 95% confidence intervals (CIs) from 3 models: Crude Model (unadjusted); Model 1 (adjusted for age, sex, race, education, PIR, BMI, marital status, smoking status, alcohol consumption, physical activity, and HEI); and Model 2 (further adjusted for hypertension, hyperlipidemia, diabetes, and cancer). The curve on the right, analyzed using a restricted cubic spline (RCS), illustrates the nonlinear dose-response relationship; the solid line represents the estimated association, and the shaded area is the 95% CI. BMI = body mass index, HEI = Healthy Eating Index, PIR = poverty income ratio.
The dose-response analysis confirmed a significant nonlinear, U-shaped relationship between sleep duration and PhenoAgeAccel (P for nonlinear = 0.0001), with an inflection point at approximately 6.774 hours. This suggests that both insufficient and excessive sleep may be linked to accelerated biological aging.
3.3. Association analysis between sleep time and PhenoAgeAccel in American adults (gender subgroups)
In the male subgroup (Fig. 4), the analysis revealed a similar pattern. While a significant protective effect for the 7 to 9 hour sleep group was observed in the crude model (β = −0.917; P < .001), this effect did not retain statistical significance after multivariate adjustments in Model 1 and Model 2. Conversely, sleeping more than 9 hours was associated with significantly accelerated aging in the fully adjusted Model 2 (β = 0.989; P = .044). The nonlinear curve for males showed an inflection point at 6.724 hours.
Figure 4.
Association analysis between sleep duration and PhenoAgeAccel in American adults (male). This figure displays the association between sleep duration and PhenoAgeAccel. The forest plot (left) shows the regression coefficients (β) and 95% confidence intervals (CIs) from 3 models: Crude Model (unadjusted); Model 1 (adjusted for age, race, education, PIR, BMI, marital status, smoking status, alcohol consumption, physical activity, and HEI); and Model 2 (further adjusted for hypertension, hyperlipidemia, diabetes, and cancer). The curve on the right, analyzed using a restricted cubic spline (RCS), illustrates the nonlinear dose-response relationship; the solid line represents the estimated association, and the shaded area is the 95% CI. BMI = body mass index, HEI = Healthy Eating Index, PIR = poverty income ratio.
In the female subgroup (Fig. 5), the protective association for the 7 to 9 hour sleep group appeared more pronounced. The effect was highly significant in the crude model (β = −1.445; P < .0001) and, after full adjustment in Model 2, showed a strong trend that approached statistical significance (β = −0.401; P = .091). In contrast to males, sleeping >9 hours was not significantly associated with accelerated aging in females in any of the models. The dose-response curve for females identified an inflection point at 6.623 hours.
Figure 5.
Association analysis between sleep duration and PhenoAgeAccel in American adults (female). This figure displays the association between sleep duration and PhenoAgeAccel. The forest plot (left) shows the regression coefficients (β) and 95% confidence intervals (CIs) from 3 models: Crude Model (unadjusted); Model 1 (adjusted for age, race, education, PIR, BMI, marital status, smoking status, alcohol consumption, physical activity, and HEI); and Model 2 (further adjusted for hypertension, hyperlipidemia, diabetes, and cancer). The curve on the right, analyzed using a restricted cubic spline (RCS), illustrates the nonlinear dose-response relationship; the solid line represents the estimated association, and the shaded area is the 95% CI. BMI = body mass index, HEI = Healthy Eating Index, PIR = poverty income ratio.
3.4. Interaction effect test
In this study, interaction effect tests were conducted to explore whether the relationship between sleep duration and PhenoAgeAccel was influenced by demographic and health-related characteristics such as gender, age, race, education level, economic status, BMI, marital status, smoking and drinking habits, physical activity, dietary quality, and chronic disease conditions. The results indicated that significant interaction effects were observed only among current smokers and individuals with diabetes (P < .01), with no such effects found in subgroups of other characteristics (Table 2). This finding demonstrates the robustness of the relationship between sleep duration and PhenoAgeAccel across most subgroups. For smokers and individuals with diabetes, further specific interventions may be required to elucidate potential associations.
Table 2.
Interaction effect test.
| Characteristic | <7 hours/day | 7–9 hours/day | P | >9 hours/day | P | P for trend | P for interaction |
|---|---|---|---|---|---|---|---|
| Gender | .38 | ||||||
| Female | ref | −0.4 (−0.87, 0.07) | .09 | 0.63 (−0.29, 1.55) | .17 | .94 | |
| Male | ref | −0.16 (−0.60, 0.29) | .47 | 0.99 (0.03, 1.95) | .04 | .49 | |
| Age | .31 | ||||||
| 20–29 | ref | −0.92 (−1.61, −0.22) | .01 | −0.52 (−1.45, 0.41) | .26 | .04 | |
| 30–39 | ref | 0.23 (−0.34, 0.79) | .41 | 1.07 (−0.70, 2.84) | .22 | .22 | |
| 40–49 | ref | −0.37 (−1.28, 0.53) | .4 | 0.62 (−0.86, 2.11) | .39 | .84 | |
| 50–59 | ref | −0.43 (−1.50, 0.64) | .41 | 1.16 (−1.15, 3.47) | .31 | .84 | |
| ≥60 | ref | 0.08 (−0.59, 0.74) | .81 | 1.68 (0.45, 2.91) | .01 | .05 | |
| Race | .55 | ||||||
| Non-Hispanic White | ref | −0.37 (−0.85, 0.11) | .12 | 0.9 (0.13, 1.66) | .02 | .86 | |
| Mexican American | ref | 0.26 (−0.42, 0.94) | .43 | 0.03 (−1.46, 1.52) | .97 | .59 | |
| Non-Hispanic Black | ref | −0.37 (−1.29, 0.55) | .41 | 0.87 (−1.75, 3.50) | .49 | .92 | |
| Other race | ref | 0.29 (−0.87, 1.44) | .61 | 0.76 (−1.39, 2.90) | .46 | .44 | |
| Educational level | .35 | ||||||
| Below | ref | −0.29 (−0.95, 0.38) | .37 | 0.17 (−0.87, 1.21) | .73 | .71 | |
| High school | ref | −0.67 (−1.34, −0.01) | .05 | 1.03 (−0.50, 2.57) | .17 | .79 | |
| Above | ref | −0.06 (−0.51, 0.39) | .78 | 1.06 (0.12, 1.99) | .03 | .28 | |
| Poverty to income ratio | .69 | ||||||
| <1.3 | ref | −0.19 (−0.80, 0.41) | .51 | 0.22 (−0.68, 1.11) | .61 | .88 | |
| 1.3–3.49 | ref | −0.14 (−0.73, 0.44) | .61 | 1.17 (0.21, 2.14) | .02 | .28 | |
| ≥3.5 | ref | −0.27 (−0.80, 0.27) | .31 | 1.02 (−0.11, 2.15) | .07 | .77 | |
| Body mass index | .13 | ||||||
| <25 | ref | −0.33 (−0.88, 0.22) | .22 | 0.26 (−0.71, 1.24) | .58 | .76 | |
| 25–29.9 | ref | −0.03 (−0.63, 0.58) | .93 | 0.71 (−0.44, 1.86) | .21 | .56 | |
| ≥30 | ref | −0.28 (−0.92, 0.36) | .37 | 2.1 (0.61, 3.60) | .01 | .28 | |
| Marital status | .1 | ||||||
| Widowed/divorced/separated | ref | 0.05 (−0.86, 0.95) | .92 | 2.08 (0.35, 3.82) | .02 | .14 | |
| Never married | ref | −1.18 (−1.92, −0.45) | .004 | −0.1 (−1.17, 0.97) | .84 | .06 | |
| Married/living with partner | ref | −0.13 (−0.54, 0.28) | .51 | 0.88 (0.19, 1.57) | .02 | .41 | |
| Smoke status | .04 | ||||||
| Former smoker | ref | −0.07 (−0.76, 0.63) | .84 | 1.62 (0.31, 2.93) | .02 | .27 | |
| Nonsmoker | ref | −0.19 (−0.66, 0.29) | .42 | 1.21 (0.32, 2.11) | .01 | .3 | |
| Current smoker | ref | −0.52 (−1.03, 0.00) | .05 | −0.63 (−1.64, 0.38) | .21 | .04 | |
| Alcohol status | .68 | ||||||
| Former | ref | −0.15 (−1.10, 0.79) | .73 | 1.57 (−0.75, 3.90) | .17 | .46 | |
| Never | ref | −0.59 (−1.77, 0.59) | .3 | 1.52 (−1.16, 4.19) | .25 | .68 | |
| Mild | ref | −0.25 (−0.80, 0.30) | .35 | 0.69 (−0.30, 1.69) | .16 | .96 | |
| Moderate | ref | 0.21 (−0.61, 1.02) | .6 | 0.88 (−0.28, 2.04) | .13 | .24 | |
| Heavy | ref | −0.43 (−1.06, 0.20) | .17 | 0.25 (−0.76, 1.27) | .61 | .48 | |
| Physical activity | .78 | ||||||
| Q1 | ref | −0.2 (−0.92, 0.52) | .56 | 0.93 (−0.14, 1.99) | .08 | .58 | |
| Q2 | ref | −0.4 (−1.01, 0.20) | .18 | 0.8 (−0.24, 1.84) | .12 | 1 | |
| Q3 | ref | −0.01 (−0.54, 0.52) | .96 | 0.88 (−0.63, 2.39) | .24 | .36 | |
| Healthy eating index | .67 | ||||||
| Q1 | ref | −0.3 (−0.93, 0.33) | .33 | 1.13 (−0.12, 2.37) | .07 | .64 | |
| Q2 | ref | −0.32 (−0.85, 0.20) | .21 | 0.29 (−0.90, 1.47) | .61 | .63 | |
| Q3 | ref | −0.04 (−0.51, 0.44) | .87 | 1.28 (0.19, 2.38) | .02 | .11 | |
| Diabetes | .01 | ||||||
| Yes | ref | −0.39 (−0.70, −0.08) | .02 | 0.25 (−0.48, 0.98) | .48 | .3 | |
| No | ref | 0.33 (−0.67, 1.33) | .49 | 3.33 (1.73, 4.94) | <.001 | .01 | |
| Hyperlipidemia | .45 | ||||||
| Yes | ref | −0.21 (−0.64, 0.21) | .31 | 1.19 (0.44, 1.94) | .004 | .24 | |
| No | ref | −0.33 (−0.98, 0.33) | .31 | 0.12 (−1.27, 1.51) | .86 | .62 | |
| Hypertension | .29 | ||||||
| Yes | ref | −0.15 (−0.85, 0.55) | .66 | 1.6 (0.58, 2.63) | .004 | .2 | |
| No | ref | −0.28 (−0.72, 0.17) | .21 | 0.5 (−0.34, 1.34) | .22 | .9 | |
| Cancer | .59 | ||||||
| Yes | ref | 0.64 (−0.65, 1.94) | .31 | 1.72 (−0.28, 3.72) | .09 | .09 | |
| No | ref | −0.31 (−0.67, 0.06) | .09 | 0.79 (0.04, 1.54) | .04 | .87 |
Q1 = the first quartile, Q2 = the second quartile, Q3 = the third quartile, ref = reference.
4. Discussion
This study delved into the dose-response relationship between sleep duration and PhenoAgeAccel, uncovering its nonlinear characteristics. Through a detailed analysis of NHANES data from 2001 to 2010, we discovered a significant positive association between daily sleep durations of 7 to 9 hours and deceleration in phenotypic aging, whereas sleep durations below or beyond this range were linked to accelerated phenotypic aging. This dose-response finding not only augments existing research on sleep and aging but also holds crucial implications for public health guidance and the optimization of individual sleep habits.
Particularly noteworthy is our identification of a critical inflection point (~6.7 hours/day) in the nonlinear model between sleep duration and PhenoAgeAccel, providing a more precise quantitative description of the relationship between sleep duration and biological aging. This finding underscores the pivotal role of moderate sleep in preventing the acceleration of biological aging, suggesting that both insufficient and excessive sleep could adversely affect health. Moreover, gender subgroup analysis further revealed variations in the dose-response effect across different gender groups, indicating the need to consider gender differences when devising targeted health interventions. Additionally, the results of the interaction effect tests suggest that an individual’s health status (such as smoking and diabetes status) may influence the dose-response relationship between sleep duration and PhenoAgeAccel, offering valuable insights for personalized medicine.
In delving deeper into the relationship between sleep and biological aging, our focus extends beyond the duration of sleep to understand the profound impact of sleep quality on physiological mechanisms. Recent studies have unveiled the intricate interplay between sleep and aging, providing new perspectives on this relationship.
Firstly, sleep is crucial for maintaining the stability and functionality of the nervous system. Research by Mander et al[15] demonstrated that adequate sleep facilitates the clearance of metabolic waste from the brain, including beta-amyloid proteins associated with Alzheimer disease, primarily during deep sleep stages through the so-called “glymphatic system,” aiding in brain health maintenance and cognitive function preservation. Secondly, sleep is closely linked to the body’s endocrine system. Studies by Knutson et al[16] have shown that insufficient or poor-quality sleep disrupts the normal secretion of hormones, including growth hormone, cortisol, and insulin. Hormonal imbalances not only affect energy metabolism but may also accelerate cellular aging and tissue damage, thereby promoting the process of biological aging.
Furthermore, sleep is intimately connected with immune system function. Research by Irwin[17] has shown that good sleep helps maintain immune system balance and enhances the body’s resistance to inflammation and infection. Conversely, sleep deprivation leads to increased inflammatory responses, which, over time, may accelerate the aging of cells and increase the risk of chronic diseases. Studies by Besedovsky et al[18] point out a bidirectional link between sleep and immunity. Activation of the immune system alters sleep patterns, which in turn affects the innate and adaptive parts of our defense system. Inflammatory responses induced by microbial challenges, depending on their magnitude and temporal course, can induce increases in sleep duration and intensity but may also lead to sleep disruption. Enhanced sleep during infection is considered a way to feedback to the immune system to promote host defense. In fact, sleep influences various immune parameters, associated with reduced infection risk, and can improve infection outcomes and vaccine responses. A possible mechanism supporting the immune-supportive effects of sleep is the induction of a hormone combination supportive of immune function. Moreover, research by Ibarra-Coronado et al[19] suggests sleep is a significant regulator of immune responses. Thus, lack of sleep weakens immunity, increasing susceptibility to infections. For instance, shorter sleep durations are associated with increased risk of the common cold. The role of sleep in altering immune responses needs to be determined to understand how sleep deprivation increases susceptibility to viral, bacterial, and parasitic infections. Several explanations for increased susceptibility to infections after sleep deprivation, such as impaired lymphocyte mitotic proliferation, reduced HLA-DR expression, upregulation of CD14+, and alterations in CD4+ and CD8+ T lymphocytes, have been observed during partial sleep deprivation.
This study echoes the findings of Hahn et al, who discovered in a mouse model that adequate sleep significantly increased the number and functionality of classical monocytes in circulation and enhanced defense outcomes against bacterial infections, thereby improving survival rates. This suggests that sleep enhances the body’s early defense against infections by boosting the quantity and functionality of immune cells, possibly through the regulation of hormone levels and cytokine production related to the immune system. Moreover, the research by Christoffersson et al revealed the impact of acute sleep deprivation on neutrophils, the most abundant cell type in our immune system and the first line of defense against infection responses, in healthy young males. They found that sleep deprivation led to an increase in neutrophil count but impaired their functionality, such as reduced production of reactive oxygen species, potentially lowering the body’s defense capability against infections. These studies underscore the importance of sleep for biological health, especially in maintaining the balance and functionality of the immune system. While this research focuses on the relationship between sleep and biological aging, these studies provide insights into the effects of sleep on immune cells, indirectly supporting the crucial role of sleep in maintaining overall biological health, including delaying the aging process. These findings emphasize the importance of maintaining proper sleep patterns for promoting health and preventing disease, complementing the objectives of this study.
This research delved into the dose-response relationship between sleep duration and PhenoAgeAccel, revealing its nonlinear characteristics and finding that moderate sleep per day is significantly associated with a deceleration in the biological aging process. This discovery not only enriches existing research in the field of sleep and aging but also provides crucial scientific evidence for public health guidance and the optimization of individual sleep habits. The importance of moderate sleep extends beyond its role in regulating the immune system; it plays a key role in maintaining the balance of various physiological systems in the body. For instance, adequate sleep facilitates the clearance of metabolic waste from the brain, reduces the accumulation of proteins associated with neurodegenerative diseases,[20–23] and protects cognitive functions from damage. Additionally, moderate sleep balances the endocrine system, regulates hormone levels, and plays a vital role in emotional stability and metabolic health.[24–26] From the perspective of biological aging, moderate sleep supports the slowing of the aging process through various mechanisms, such as maintaining the effectiveness of DNA repair mechanisms, alleviating oxidative stress and inflammation, and influencing telomere length and cellular aging pathways. The combined effect of these mechanisms not only slows the accumulation of cellular damage and genetic mutations but also helps maintain the youthful state of cells and tissues, thereby extending healthy lifespan.[27–31] Furthermore, the interaction effects observed in this study for smoking and diabetes conditions may interact through shared physiological pathways, affecting sleep patterns and biological aging. For example, both smoking and diabetes are associated with endocrine changes that may disrupt sleep regulation mechanisms, leading to decreased sleep quality and changes in sleep duration, which in turn affect biological aging.
Therefore, moderate sleep should be viewed as an important public health strategy, capable of not only improving current quality of life but also laying the foundation for future health and longevity. Future research should delve deeper into the intrinsic connections between sleep and biological aging, exploring the specific mechanisms through which improved sleep quality can slow the aging process, providing a solid scientific basis for effective health interventions.
Despite providing valuable insights into the relationship between sleep duration and PhenoAgeAccel, this study has several limitations to note. First, due to its cross-sectional study design, causality cannot be established. Although we observed a significant association between sleep duration and PhenoAgeAccel, we cannot determine whether sleep patterns influence the biological aging process or vice versa. Second, sleep duration data was based on participants’ self-reports, which may be subject to recall bias. Self-reported sleep duration may not be as accurate as objective sleep monitoring (such as polysomnography), potentially affecting our assessment of the relationship between sleep duration and PhenoAgeAccel. Furthermore, we did not thoroughly consider the impact of sleep quality, including sleep interruptions and depth, which are important factors affecting biological aging. Although we controlled for various potential confounders, including lifestyle, health status, and socioeconomic status, there may still be unobserved confounders. For example, genetic factors, long-term illness history, medication use, and environmental factors may also affect sleep patterns and the biological aging process, but these factors were not fully considered in the current study. Lastly, the sample of this study was drawn from the NHANES in the United States, and while representative, the results may not be entirely applicable to other countries and regions. Different cultural backgrounds, lifestyles, and social environments may influence the relationship between sleep patterns and biological aging. Despite these limitations, the findings of this study provide valuable insights into the relationship between sleep and biological aging, offering important references for future research directions and public health interventions. Future research should adopt longitudinal study designs, use objective sleep monitoring tools, and consider more potential influencing factors to further elucidate the causal relationships and mechanisms between sleep and biological aging.
5. Conclusion
This study suggests that moderate sleep duration of 7 to 9 hours/day is associated with a deceleration in phenotypic aging, with a critical inflection point for sleep and aging health at approximately 6.7 hours. Both insufficient and excessive sleep durations may be detrimental to slowing the aging process. The results of the interaction effect tests emphasize the need to consider individual smoking and diabetes status when developing targeted health interventions. These findings provide new insights into the complex relationship between sleep and biological aging, offering a scientific basis for public health guidance and the optimization of individual sleep habits. Future research should employ longitudinal designs and objective sleep monitoring tools to further explore the causal relationship and underlying mechanisms between sleep and biological aging.
Acknowledgments
The authors express their gratitude to the National Health and Nutrition Examination Survey (NHANES) for providing the public access data for this study. We also sincerely thank all the NHANES participants for their invaluable contributions, which made this research possible.
Author contributions
Conceptualization: Yujia Liu, Kai Qiao.
Data curation: Yujia Liu.
Formal analysis: Yujia Liu, Kai Qiao.
Methodology: Yujia Liu.
Visualization: Yujia Liu.
Writing – original draft: Kai Qiao.
Writing – review & editing: Kai Qiao.
Abbreviations:
- BMI
- body mass index
- HEI-2020
- Healthy Eating Index-2020
- NHANES
- National Health and Nutrition Examination Survey
- PhenoAgeAccel
- phenotypic age acceleration
- PIR
- poverty income ratio
This study was a secondary analysis based on data from the National Health and Nutrition Examination Survey (NHANES) for the cycles 2001 to 2010. The survey protocols for NHANES were approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board (ERB). Specifically, the data cycles from 2001 to 2004 were conducted under Protocol #98-12, and the cycles from 2005 to 2010 were conducted under Protocol #2005-06. All participants provided written informed consent. As this study used only publicly available and de-identified data, it was exempt from further institutional review board approval.
The authors have no funding and conflicts of interest to disclose.
The datasets generated during and/or analyzed during the current study are publicly available.
How to cite this article: Liu Y, Qiao K. Dose-response relationship between sleep duration and mediation of phenotypic age acceleration: A cross-sectional study. Medicine 2025;104:40(e44786).
References
- [1].Ward-Caviness CK, Russell AG, Weaver AM, et al. Accelerated epigenetic age as a biomarker of cardiovascular sensitivity to traffic-related air pollution. Aging (Albany NY). 2020;12:24141–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Chen L, Yin X, Zhao Y, et al. Biological ageing and the risks of all-cause and cause-specific mortality among people with diabetes: a prospective cohort study. J Epidemiol Community Health. 2022;76:771–8. [DOI] [PubMed] [Google Scholar]
- [3].Shadyab AH, McEvoy LK, Horvath S, et al. Association of epigenetic age acceleration with incident mild cognitive impairment and dementia among older women. J Gerontol A Biol Sci Med Sci. 2021;77:1239–44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Raniti MB, Allen NB, Schwartz O, et al. Sleep duration and sleep quality: associations with depressive symptoms across adolescence. Behav Sleep Med. 2017;15:198–215. [DOI] [PubMed] [Google Scholar]
- [5].Chee MWL. Fragmented sleep and cortical thinning in old adults: time to wake up? Sleep. 2016;39:15–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Zhi TF, Sun XM, Li SJ, et al. Associations of sleep duration and sleep quality with life satisfaction in elderly Chinese: the mediating role of depression. Arch Gerontol Geriatr. 2016;65:211–7. [DOI] [PubMed] [Google Scholar]
- [7].Lo JC, Groeger JA, Cheng GH, Dijk DJ, Chee MWL. Self-reported sleep duration and cognitive performance in older adults: a systematic review and meta-analysis. Sleep Med. 2016;17:87–98. [DOI] [PubMed] [Google Scholar]
- [8].Levine M, Lu A, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY). 2018;10:573–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Wu D, Huang P, Geng X, et al. Dose-response relationship between leisure-time physical activity patterns and phenotypic age acceleration in American adults: a cross-sectional analysis. J Exerc Sci Fit. 2024;22:445–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Wu D, Jia Y, Liu Y, Shang M. Dose-response relationship of dietary omega-3 fatty acids on slowing phenotypic age acceleration: a cross-sectional study. Front Nutr. 2024;11:1424156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Liu Z, Kuo PL, Horvath S, Crimmins E, Ferrucci L, Levine M. A new aging measure captures morbidity and mortality risk across diverse subpopulations from NHANES IV: a cohort study. PLoS Med. 2018;15:e1002718. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Wu D, Shen Y, Qu C, et al. Association between dietary and behavioral-based oxidative balance score and phenotypic age acceleration: a cross-sectional study of Americans. Epidemiol Health. 2024;46:e2024023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Wu D, Jia Y, Liu Y, Pan X, Li P, Shang M. Dose response of leisure time physical activity and biological aging in type 2 diabetes: a cross sectional study. Sci Rep. 2024;14:26253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Reedy J, Pannucci T, Herrick K, Lerman J, Shams-White M, Zimmer M. Healthy Eating Index protocol: review, update, and development process to reflect dietary guidance across the lifespan. Curr Dev Nutr. 2021;5:447. [Google Scholar]
- [15].Mander BA, Winer JR, Walker MP. Sleep and human aging. Neuron. 2017;94:19–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Knutson KL, Spiegel K, Penev P, Van Cauter E. The metabolic consequences of sleep deprivation. Sleep Med Rev. 2007;11:163–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Irwin MR. Sleep and inflammation: partners in sickness and in health. Nat Rev Immunol. 2019;19:702–15. [DOI] [PubMed] [Google Scholar]
- [18].Besedovsky L, Lange T, Haack M. The sleep-immune crosstalk in health and disease. Physiol Rev. 2019;99:1325–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Ibarra-Coronado EG, Pantaleón-Martínez AM, Velazquéz-Moctezuma J, et al. The bidirectional relationship between sleep and immunity against infections. J Immunol Res. 2015;2015:678164. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Morrone CD, Raghuraman R, Hussaini SA, Yu WH. Proteostasis failure exacerbates neuronal circuit dysfunction and sleep impairments in Alzheimer’s disease. Mol Neurodegener. 2023;18:27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Hafycz JM, Naidoo NN. Sleep, aging, and cellular health: aged-related changes in sleep and protein homeostasis converge in neurodegenerative diseases. Front Aging Neurosci. 2019;11:140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Picchioni D, Schmidt KC, Loutaev I, et al. Increased rates of brain protein synthesis during [N1,N2] sleep: L-[1-11C]leucine PET studies in human subjects. J Cereb Blood Flow Metab. 2023;43:59–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Tsiknia AA, Parada H, Banks SJ, Reas ET. Sleep quality and sleep duration predict brain microstructure among community-dwelling older adults. Neurobiol Aging. 2023;125:90–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Palamarchuk IS, Slavich GM, Vaillancourt T, Rajji TK. Stress-related cellular pathophysiology as a crosstalk risk factor for neurocognitive and psychiatric disorders. BMC Neurosci. 2023;24:65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Luo F, Sandhu AF, Rungratanawanich W, et al. Melatonin and autophagy in aging-related neurodegenerative diseases. Int J Mol Sci. 2020;21:7174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Faggiano A, Gherbesi E, Avagimyan A, et al. Melatonin mitigates oxidative damage induced by anthracycline: a systematic-review and meta-analysis of murine models. Front Cardiovasc Med. 2023;10:1289384. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Mak MHC, O’Hagan A, Horner AJ, Gaskell MG. A registered report testing the effect of sleep on Deese-Roediger-McDermott false memory: greater lure and veridical recall but fewer intrusions after sleep. R Soc Open Sci. 2023;10:220595. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Sato M, Betriana F, Tanioka R, Osaka K, Tanioka T, Schoenhofer S. Balance of autonomic nervous activity, exercise, and sleep status in older adults: a review of the literature. Int J Environ Res Public Health. 2021;18:12896. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Scullin MK, Bliwise DL. Sleep, cognition, and normal aging: integrating a half century of multidisciplinary research. Perspect Psychol Sci. 2015;10:97–137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Chery MJ, Baral A, Rolle LD, et al. Depression, sleep health & sociodemographic correlates in a nationwide survey: implications for depression treatment during the COVID-19. Nat Sci Sleep. 2024;16:17–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Liu X, Xu P, Wei R, et al. Gender- and age-specific associations of sleep duration and quality with cognitive impairment in community-dwelling older adults in Anhui Province, China. Front Public Health. 2023;11:1047025. [DOI] [PMC free article] [PubMed] [Google Scholar]