Physiological Levels of Melatonin Relate to Cognitive Function and Depressive Symptoms: The HEIJO-KYO Cohort

Context:

In contrast with randomized controlled trials, observational studies have suggested that physiological levels of melatonin are reduced in patients with dementia or depression, but the relationship has not been evaluated in large populations.

Objective:

The objective was to determine the relationships between physiological levels of melatonin and cognitive function and depressive symptoms.

Design and Participants:

A cohort of 1105 community-dwelling elderly individuals was enrolled in this cross-sectional study (mean age, 71.8 ± 7.1 y).

Measures:

Urinary 6-sulfatoxymelatonin excretion (UME) and Mini-Mental State Examination (MMSE; n = 935) and Geriatric Depression Scale (GDS; n = 1097) scores were measured as indices of physiological melatonin levels, cognitive function, and depressive symptoms, respectively.

Results:

With increases in UME quartiles, the prevalence of cognitive impairment (MMSE score ≤ 26) and depressed mood (GDS score ≥ 6) significantly decreased (P for trend = .003 and .012, respectively). In multivariate logistic regression models, after adjusting for confounders such as age, gender, socioeconomic status, physical activity, and sleep/wake cycles, higher UME levels were significantly associated with lower odds ratios (ORs) for cognitive impairment and depressed mood (ORs: Q1 = 1.00; Q2 = 0.88 and 0.76; Q3 = 0.66 and 0.85; Q4 = 0.67 and 0.53; P for trend = .023 and .033, respectively). In addition, the highest UME group showed a significantly lower OR for depressed mood than the lowest UME group (Q4 vs Q1: OR, 0.53; 95% confidence interval, 0.32–0.89; P = .033). UME levels above the median value were significantly associated with a lower OR for cognitive impairment, even after further adjustment for depressive symptoms (OR = 0.74; 95% confidence interval, 0.55–0.99; P = .043).

Conclusions:

Significant associations of higher physiological melatonin levels with lower prevalence of cognitive impairment and depressed mood were revealed in a large general elderly population. The association between physiological melatonin levels and cognitive function was independent of depressive symptoms.

A gradual cognitive decline is observed in most elderly people. Population-based studies suggest that, in developed countries, mild cognitive impairment (MCI), which is not reaching dementia diagnosis, is more prevalent (10–20%) than dementia (5–6%) (1–3). Annually, 10–15% of elderly people with MCI progress to dementia compared with 1–2% of elderly people with normal cognition (4–5). MCI is also associated with functional impairment, decreased quality of life, and early mortality (6–8). Therefore, measures to prevent cognitive impairment in the aging population are important; however, the risk factors for cognitive impairment are not fully understood.

The prevalence of depression has increased in recent decades, and depression is an important modifiable risk factor for dementia caused by Alzheimer's disease, vascular dementia, and psychiatric dementia (9, 10). Depression is also a risk factor for MCI (11). Therefore, preventing depression in elderly people could decrease cognitive impairment.

Accumulating evidence suggests that misalignment in circadian biological rhythmicity may predispose to dementia and depression (12, 13). Melatonin, a pineal gland hormone, is a biomarker of circadian biological rhythmicity produced mostly at night. Decreased physiological melatonin levels have been reported in individuals with circadian misalignment (14). Melatonin is also a potent antioxidant that may protect against atherosclerosis by lowering blood pressure and regulating sleep and biological rhythms (14, 15). To date, no significant effects of melatonin administration on cognition or mood have been identified in randomized controlled trials (16, 17). In contrast, observational studies suggest that physiological levels of melatonin are reduced in demented or depressed patients (18–23). However, these observational studies have been limited by small numbers of participants.

In this cross-sectional study involving a large cohort of 1105 community-dwelling elderly people, we investigated the relationships between physiological levels of melatonin and cognitive function and depressive symptoms.

Subjects and Methods

Participants

Between September 2010 and April 2014, 1127 community-dwelling elderly subjects (age ≥ 60 y) were voluntarily enrolled in the Housing Environments and Health Investigation among Japanese Older People in Nara, Kansai Region, a prospective community-based cohort study (HEIJO-KYO) (24). We excluded participants from this study who did not complete measurements of physiological melatonin levels and cognitive function or mood; 1105 participants remained. All participants provided written informed consent, and the study protocol was approved by the medical ethics committee of Nara Medical University.

Measurement of cognitive function

Trained clinical psychologists assessed cognitive function using the Mini-Mental State Examination (MMSE); higher MMSE scores indicate better cognitive function (range, 0–30). In this study, scores ≤ 26 indicated cognitive impairment. This higher cutoff value provides a better balance of sensitivity and specificity than the traditional value of 23 in an educated population (0.69 and 0.91 vs 0.45 and 1.00, respectively, for cognitive impairment; and 0.79 and 0.90 vs 0.58 and 0.98, respectively, for Alzheimer's disease) (25, 26).

Measurement of depressive symptoms

Depressive symptoms were measured using the short version of the Geriatric Depression Scale (GDS) (GDS-15). Scores ≥ 6 indicated depressed mood. Previous studies validating this cutoff value reported a sensitivity and specificity for clinically diagnosed depression of 0.85–0.91 and 0.68–0.78, respectively (27). Some studies suggested that the cutoff value of 6 was optimal, although a cutoff value of 5, with higher sensitivity but lower specificity, was used in our previous study (28).

Urinary 6-sulfatoxymelatonin excretion (UME) measurement

Participants were instructed to collect their urine overnight. The last void at bedtime was discarded and then each subsequent void was collected, including the first morning void. Samples were stored in a dark bottle with a cold pack at room temperature until the next day. The total volume was measured, and samples were stored at −20°C until analysis. Urinary 6-sulfatoxymelatonin levels were measured at a commercial laboratory (SRL, Inc) using an ELISA kit (RE54031; IBL International) with high sensitivity. UME was calculated using the following formula: UME (μg) = 6-sulfatoxymelatonin level (μg/mL) × total overnight urine volume (mL). If urine was not collected as per protocol, UME data were considered missing. UME reproducibility was evaluated in the first 188 participants, and the intraclass correlation coefficient over approximately 4 months was moderate (29).

Other measurements

Educational level, household income, history of night-shift work, alcohol consumption, and information on medicines were evaluated using a self-administered questionnaire. Hypertension was documented based on medical history and current antihypertensive therapy. Diabetes mellitus was documented based on medical history, current antidiabetic therapy, and fasting plasma glucose and glycated hemoglobin levels. Daytime physical activity was measured at 1-minute intervals during waking hours using an actigraph (Actiwatch 2; Respironics Inc) worn on the nondominant wrist. The average of physical activity counts was automatically calculated by Actiware version 5.5 (Respironics) with the default algorithm. Bedtime and duration in bed were measured using a self-reported sleep diary. Day-to-day correlations of daytime physical activity (r = 0.83), bedtime (r = 0.80), and duration in bed (r = 0.72) were moderately high. Data on day length in Nara (latitude 34°N) from sunrise to sunset on the days of measurement were available in the national database. Nighttime bedroom light intensity was measured at 1-minute intervals using a light meter (LX-28SD; Sato Shouji Inc) with the sensor fixed 60 cm above the floor near the head of the bed and facing the ceiling. Regarding data on physical activity, bedtime, duration in bed, day length, and nighttime bedroom light intensity, the average of data collected on 2 consecutive days before UME measurement was used for analysis.

Statistical analysis

Means and medians between normal and impaired cognitive function groups and normal and depressed mood groups were compared using the unpaired t test and Mann-Whitney U test, respectively. The χ² test was used for comparison of categorical data. Association trends of UME quartiles with medians of MMSE and GDS scores and prevalence of cognitive impairment and depressed mood were analyzed using the Jonckheere-Terpstra test for trends and logistic regression analysis, respectively. Logistic regression models included cognitive and mood status as dependent variables and UME (per quartile), age (≥ 70 y), gender, body mass index (BMI), educational level (≥ 13 y), household income (≥ 4 million Japanese yen/y), night-shift work history, alcohol consumption (≥ 30 g/d), hypertension, diabetes, daytime physical activity (per 100 counts/min), bedtime (per 1-h delay), duration in bed (scotoperiod; per 1 h), day length (photoperiod; per quartile increment), and nighttime bedroom light intensity (average ≥ 5 lux) as independent variables. In multivariate statistical models, model 1 was adjusted for age, model 2 was adjusted for independent variables associated with mood or cognitive status in Table 2 (P < .20), and model 3 was adjusted for all independent variables in Table 2. No serious multicollinearity was observed (all variance inflation factors < 10) in any of the multivariate models. Statistical analysis was performed using SPSS version 19.0 for Windows (IBM SPSS Inc). A two-sided P value <.05 was considered statistically significant.

Table 2.

Basic and Circadian Rhythm Parameters by Cognitive and Mood Status

Characteristics	Cognitive Function		P	Mood		P
	Impaired (MMSE ≤ 26)	Preserved (MMSE ≥ 27)		Depressed (GDS ≥ 6)	Non-Depressed (GDS ≤ 5)
No. of participants	311	624		166	931
Examination score, median (range)	25 (13, 26)	29 (27, 30)	<.001	7 (6, 14)	2 (0, 5)	<.001
Basic parameters
Age, mean (SD), y	73.7 (7.0)	70.6 (6.9)	<.001	72.7 (7.6)	71.7 (7.0)	.09
Gender (male), n (%)	136 (43.7)	300 (48.1)	.21	67 (40.4)	448 (48.1)	.07
BMI, mean (SD), kg/m2	23.2 (3.2)	23.1 (3.0)	.58	23.0 (3.2)	23.1 (3.0)	.49
Education (≥13 y), n (%)	60 (19.3)	206 (33.0)	<.001	32 (19.3)	264 (28.4)	.015
Household income (≥ 4 million Japanese yen/y), n (%)	97 (35.0)	276 (46.6)	.001	53 (35.6)	388 (45.0)	.033
Night-shift work history, n (%)	28 (9.4)	38 (6.3)	.08	8 (5.4)	63 (7.3)	.42
Circadian rhythm parameters
Daytime physical activity, mean (SD), count/min	295.4 (103.8)	301.3 (103.2)	.41	285.0 (104.7)	300.1 (103.1)	.08
Bedtime, mean (SD), clock time	22:12 (1:08)	22:38 (1:10)	<.001	22:16 (1:13)	22:31 (1:10)	.015
Duration in bed (scotoperiod), mean (SD), min	514.8 (76.9)	487.5 (75.3)	<.001	511.6 (86.6)	495.5 (75.0)	.013
Day length (photoperiod), median (IQR), min	658 (630, 686)	651 (623, 681)	.10	654 (614, 681)	653 (623, 682)	.64
Nighttime light levels (average ≥ 5 lux), n (%)	63 (20.4)	109 (17.6)	.30	42 (25.5)	167 (18.1)	.027

Results

Mean (SD) age of the 1105 participants was 71.8 (7.1) years, and 519 (47.0%) were male. Median UME was 6.8 μg (interquartile range [IQR], 4.0–10.5), and median MMSE (n = 935) and GDS (n = 1097) scores were 28 (IQR, 26–30) and 2 (IQR, 1–4), respectively. A quartile increase in UME was significantly associated with age, gender, BMI, and educational level (Table 1). There was no significant association trend between nighttime light levels and UME quartiles.

Table 1.

Basic and Circadian Rhythm Parameters by UME Quartiles

	Quartiles of UME				Ptrend
	Q1 2.8 [< 4.0]	Q2 5.4 [4.0, 6.8]	Q3 8.4 [6.8, 10.5]	Q4 14.1 [> 10.5]
No. of participants	277	276	276	276
Basic parameters
Age, mean (SD), y	73.4 (7.3)	73.0 (7.2)	71.3 (6.7)	69.0 (6.5)	<.001
Gender (male), n (%)	92 (33.2)	114 (41.3)	152 (55.1)	161 (58.3)	<.001
BMI, mean (SD), kg/m2	22.9 (3.5)	23.0 (3.1)	23.0 (2.8)	23.5 (2.8)	.023
Education (≥13 y), n (%)	53 (19.1)	69 (25.0)	87 (31.5)	89 (32.2)	<.001
Household income (≥ 4 million Japanese yen/y), n (%)	107 (42.0)	93 (37.7)	125 (49.0)	119 (45.2)	.13
Night-shift work history, n (%)	18 (7.2)	11 (4.4)	17 (6.6)	26 (9.8)	.15
Circadian rhythm parameters
Daytime physical activity, mean (SD), count/min	302.3 (101.6)	300.1 (105.4)	293.6 (102.6)	297.0 (107.3)	.42
Bedtime, mean (SD), clock time	22:30 (1:10)	22:27 (1:13)	22:28 (1:09)	22:29 (1:11)	.86
Duration in bed (scotoperiod), mean (SD), min	493.7 (76.9)	495.3 (82.7)	503.4 (78.8)	498.1 (70.3)	.31
Day length (photoperiod), median (IQR), min	647 (609, 682)	657 (636, 682)	651 (628, 684)	653 (623, 688.6)	.25
Nighttime light levels (average ≥ 5 lux), n (%)	52 (18.8)	52 (19.0)	58 (21.1)	51 (18.7)	.88

UME quartile values are expressed as median [IQR] in micrograms.

As shown in Table 2, the impaired cognitive function group (MMSE score ≤ 26; n = 311) showed significantly older age, lower education level, lower household income, earlier bedtime, and longer duration in bed than the preserved cognitive function group (MMSE score ≥ 27; n = 624). The depressed mood group (GDS score ≥ 6; n = 166) showed significantly lower educational level, lower household income, lower physical activity, earlier bedtime, longer duration in bed, and higher nighttime light intensity than the nondepressed mood group (GDS score ≤ 5; n = 931). Nocturnally collected urine volume did not differ significantly between the impaired and preserved cognitive function groups (571.5 vs 575.1 mL, respectively; P = .84).

With a quartile increase in UME, prevalence of impaired cognitive function and depressed mood significantly decreased (Q1, 40.0 and 19.0%; Q2, 36.1 and 15.0%; Q3, 29.1 and 15.9%; Q4, 28.3 and 10.6%; P for trend = .003 and .012, respectively; Figure 1), and median MMSE and GDS scores significantly increased and decreased, respectively (Q1, 27 and 3; Q2, 28 and 2; Q3, 28 and 2; Q4, 29 and 2; P for trend < .001 and = .001, respectively). In sensitivity analysis excluding participants taking antidepressants (n = 19) or benzodiazepines (n = 103), or with a low MMSE score ≤ 23 (n = 35), these association trends were maintained (cognitive function: P for trend = .005, .010, and .009, respectively; mood: P for trend = .006, .024, and .010, respectively).

Figure 1.

Associations of UME quartiles with prevalence of cognitive impairment and depressed mood.

Solid bars indicate the prevalence of cognitive impairment (A) and depressed mood (B). P values are shown for the trends determined by logistic regression analysis.

In univariate logistic regression analysis, with a quartile increase in UME, odds ratios (ORs) for impaired cognitive function and depressed mood significantly decreased (Q1, 1.00; Q2, 0.85 and 0.75; Q3, 0.62 and 0.81; Q4, 0.59 and 0.50; P for trend = .003 and .012, respectively; Table 3). In multivariate logistic regression models after adjusting for confounding factors, higher UME levels were consistently significantly associated with lower ORs for impaired cognitive function and depressed mood (model 1: Q1, 1.00; Q2, 0.88 and 0.75; Q3, 0.66 and 0.81; Q4, 0.69 and 0.51; P for trend = .026 and .015; model 2: Q1, 1.00; Q2, 0.88 and 0.76; Q3, 0.66 and 0.85; Q4, 0.67 and 0.53; P for trend = .023 and .033; model 3: Q1, 1.00; Q2, 0.89 and 0.75; Q3, 0.69 and 0.85; Q4, 0.70 and 0.54; P for trend = .048 and .036, respectively). Furthermore, these trends remained significant in the model adjusted for variables in model 2 plus alcohol consumption, hypertension, and diabetes (P for trend = .025 and .047 for cognitive function and depressed mood, respectively). Furthermore, the highest UME group showed a significantly lower OR for depressed mood than the lowest UME group (Q4 vs Q1, model 1: OR, 0.51; 95% confidence interval [CI], 0.31–0.83; P = .015; model 2: OR, 0.53; 95% CI, 0.32–0.89; P = .033; model 3: OR, 0.54; 95% CI, 0.32–0.89; P = .036).

Table 3.

Logistic Regression Analysis for the Associations of UME Quartiles With Cognitive Function and Mood

	Quartiles of UME				Ptrend
	Q1 2.8 [< 4.0]	Q2 5.4 [4.0, 6.8]	Q3 8.4 [6.8, 10.5]	Q4 14.1 [> 10.5]
Impaired cognitive function
Unadjusted OR (95% CI)	1.00 (ref)	0.85 (0.58, 1.23)	0.62 (0.42, 0.91)	0.59 (0.40, 0.87)	.003
Adjusted OR (95% CI)
Model 1	1.00 (ref)	0.88 (0.60, 1.29)	0.66 (0.44, 0.97)	0.69 (0.46, 1.02)	.026
Model 2	1.00 (ref)	0.88 (0.59, 1.30)	0.66 (0.44, 0.98)	0.67 (0.45, 1.01)	.023
Model 3	1.00 (ref)	0.89 (0.60, 1.32)	0.69 (0.46, 1.04)	0.70 (0.46, 1.06)	.048
Depressed mood
Unadjusted OR (95% CI)	1.00 (ref)	0.75 (0.48, 1.17)	0.81 (0.52, 1.25)	0.50 (0.31, 0.82)	.012
Adjusted OR (95% CI)
Model 1	1.00 (ref)	0.75 (0.48, 1.17)	0.81 (0.52, 1.26)	0.51 (0.31, 0.83)	.015
Model 2	1.00 (ref)	0.76 (0.48, 1.19)	0.85 (0.54, 1.35)	0.53 (0.32, 0.89)	.033
Model 3	1.00 (ref)	0.75 (0.47, 1.18)	0.85 (0.54, 1.34)	0.54 (0.32, 0.89)	.036

UME quartile values are expressed as median [IQR] in micrograms. Model 1, Adjusted for age. Model 2, Adjusted for variables associated with impaired cognitive function (age, education, household income, night-shift work, bedtime, duration in bed, and day length) or depressed mood (age, gender, education, household income, daytime physical activity, bedtime, duration in bed, and nighttime light levels) in Table 2 (P < .20). Model 3, Adjusted for all variables shown in Table 2 (age, gender, BMI, education, household income, night-shift work, daytime physical activity, bedtime, duration in bed, day length, and nighttime light levels).

When participants were divided into two groups by the median value of UME, the high UME group (n = 477) had a significantly lower OR for impaired cognitive function than the low UME group (n = 458; OR, 0.66; 95% CI, 0.50–0.87; P = .003; Table 4). In multivariate logistic regression models after adjusting for confounding factors, the high UME group had consistently and significantly lower ORs for impaired cognitive function (model 1: OR, 0.72; 95% CI, 0.54–0.95; P = .020; model 2: OR, 0.71; 95% CI, 0.53–0.95; P = .019; model 3: OR, 0.74; 95% CI, 0.55–0.99; P = .042). Even after further adjustment for mood status in model 3, higher UME levels were significantly associated with a lower OR for impaired cognitive function (model 4: OR, 0.74; 95% CI, 0.55–0.99; P = .043).

Table 4.

Logistic Regression Analysis for the Association Between UME Halves and Cognitive Function

	Halves of UME		P
	Low 4.0 [<6.8]	High 10.5 [>6.8]
No. of participants	458	477
Impaired cognitive function
Unadjusted OR (95% CI)	1.00 (ref)	0.66 (0.50, 0.87)	.003
Adjusted OR (95% CI)
Model 1	1.00 (ref)	0.72 (0.54, 0.95)	.020
Model 2	1.00 (ref)	0.71 (0.53, 0.95)	.019
Model 3	1.00 (ref)	0.74 (0.55, 0.99)	.042
Model 4	1.00 (ref)	0.74 (0.55, 0.99)	.043

Halves of UME values are expressed as median [IQR] in micrograms. Model 1, Adjusted for age. Model 2, Adjusted for variables associated with impaired cognitive function (age, education, household income, night-shift work, bedtime, duration in bed, and day length) in Table 2 (P < .20). Model 3, Adjusted for all variables shown in Table 2 (age, gender, BMI, education, household income, night-shift work, daytime physical activity, bedtime, duration in bed, day length, and nighttime light levels). Model 4, Adjusted for variables in model 3 plus mood status.

Discussion

In a large general elderly population, the prevalence of cognitive impairment and depressed mood decreased with increases in physiological melatonin levels. The association between physiological melatonin levels and cognitive function was independent of depressive symptoms and the major confounders including age, gender, socioeconomic status, physical activity, and sleep/wake cycles.

These findings are inconsistent with those of randomized controlled trials. A Cochrane systematic review, including a meta-analysis of three randomized controlled trials, found a nonsignificant effect of melatonin administration on cognition measured using the MMSE (16). In these trials, the daily dosage of melatonin was 2.5–10 mg, approximately 10- to 40-fold higher melatonin levels than physiological levels, when prolonged-release tablets were used (30). Fast-release tablets used in some studies may shorten the effective period of melatonin. In addition, the interventional period in two of the three studies was a few weeks, possibly insufficient to detect a significant effect of melatonin on cognition. Another systematic review, including a meta-analysis of 10 randomized controlled trials, concluded that there was no significant effect of melatonin against depression (17). A 3.5-year follow-up study indicated an adverse effect of 2.5 mg melatonin administration on mood (31). The differences between the previous and present results suggest that the effective range of melatonin on cognitive function and mood may exist within physiological levels. In our study, higher melatonin levels within the physiological range were associated with lower prevalence of cognitive impairment and depressed mood. Long-term randomized controlled trials testing the effects of very low-dose melatonin on cognitive function and mood should be considered.

Our study added novel findings on previous evidence reported in smaller-scale observational studies. In contrast with randomized controlled trials, our findings were generally consistent with those of previous smaller-scale observational studies reporting that physiological melatonin levels were significantly reduced in patients with Alzheimer's disease or major depressive disorder (18–23). Our study revealed that physiological melatonin levels were associated with cognition and mood even in a large general elderly population. Furthermore, depressive symptoms are an important risk factor for dementia associated with Alzheimer's disease, vascular dementia, and psychiatric dementia (9). Our study also revealed that the association between physiological melatonin levels and cognitive function was independent of depressive symptoms, suggesting that physiological melatonin levels may contribute to cognitive function through a different pathway from that affecting mood.

The possible mechanisms underlying the association between physiological levels of melatonin and cognitive function include melatonin's antioxidative action, antiamyloidogenic properties, and antidepressive and atheroprotective effects. Oxidative injury to the brain may contribute to cognitive decline (32). Melatonin acts as a potent antioxidant by free radical scavenging at extra- and intracellular levels and may have a protective effect against oxidative injury to the brain and contribute to neuroprotection against β-amyloid neurotoxicity (14, 33).

Physiological levels of melatonin influence sleep quality and circadian biological rhythmicity, and poor sleep quality and circadian misalignment are associated with metabolic abnormalities and depressed mood (14, 34, 35). Low physiological levels of melatonin are also associated with increased incidence of diabetes and hypertension (36, 37). In our previous studies, physiological melatonin levels were inversely associated with aortic stiffness, a marker of atherosclerosis, and nighttime blood pressure (29, 38). These atheroprotective effects of melatonin may also contribute to maintaining cognitive function. However, potential biological effects of physiological levels of melatonin have not been thoroughly evaluated.

Strengths of the present study include the large sample size, which enabled us to construct statistical models, including multiple potential confounders. The enhanced statistical power of this large-scale study also detected a significant association between physiological melatonin levels and depressive symptoms, which was not statistically significant in our previous study with a smaller sample size (n = 516) (28).

There are several potential limitations of this study. First, the cross-sectional design limited inference of causality for the associations of higher physiological melatonin levels with lower prevalence of cognitive impairment and depressed mood. Second, the lack of participants with clinically diagnosed dementia and depression may have led to the misclassification of cognitive and mood status. However, the moderately high agreement in previous validation studies suggests infrequent misclassification. Third, participants were not randomly selected from the general population. However, some basic data, such as BMI and estimated glomerular filtration rate reported in our previous study, were similar to the national data for Japan (29). Fourth, although statistical models included several potential confounding factors, there may be other factors that contain melatonin and influence melatonin production and metabolism. We additionally observed that the associations of physiological melatonin levels with cognitive function and mood were independent of benzodiazepine use, β-blocker use, and estimated glomerular filtration rate (data not shown). However, we could not exclude the existence of other confounding factors. Finally, physiological melatonin levels were calculated from urine samples self-collected by all participants, including those with moderate-to-severe cognitive impairment. However, the results of sensitivity analysis excluding individuals with low cognition were consistent, and urine volume did not differ between the impaired and preserved cognitive function groups.

In conclusion, in a large general elderly population, the prevalence of cognitive impairment and depressed mood decreased with increases in physiological melatonin levels. The association between physiological melatonin levels and cognitive function was independent of depressive symptoms and the major confounders including age, gender, socioeconomic status, physical activity, and sleep/wake cycles.

Acknowledgments

We thank Sachiko Uemura, Naomi Takenaka, and Keiko Nakajima for their valuable support during the collection of data.

This work was supported by research funding from the Department of Indoor Environmental Medicine, Nara Medical University; JSPS KAKENHI Grants (24790774, 22790567, 25860447, 25461393); Mitsui Sumitomo Insurance Welfare Foundation; Meiji Yasuda Life Foundation of Health and Welfare; Osaka Gas Group Welfare Foundation; Japan Diabetes Foundation; Daiwa Securities Health Foundation; the Japan Science and Technology Agency; YKK AP Inc; Nara Prefecture Health Promotion Foundation; and Nara Medical University Grant-in-Aid for Collaborative Research Projects.

Disclosure Summary: The authors report no conflicts of interest.

^*

K.O. and K.S. contributed equally to this work.

Abbreviations

 
• BMI

  body mass index

 
• CI

  confidence interval

 
• GDS

  Geriatric Depression Scale

 
• IQR

  interquartile range

 
• MCI

  mild cognitive impairment

 
• MMSE

  Mini-Mental State Examination

 
• OR

  odds ratio

 
• UME

  urinary 6-sulfatoxymelatonin excretion.