Association between serum folate and sleep duration in American adults: A cross-sectional analysis of the National Health and Nutrition Examination Survey 2005 to 2010 Authors: Yandong Dong, Mei Luo Published: 2024 ABSTRACT Folate is a vitamin that is exclusively derived from diet sources. Low folate levels are typically associated with cardiovascular disease, cognitive impairment, and cancer risk. However, few studies have examined the direct relationship between serum folate levels and sleep duration. Accordingly, we aimed to examine the relationship between serum folate levels and sleep duration. Using data from the National Health and Nutrition Examination Survey (NHANES), we examined the association between serum folate levels and sleep duration. We used laboratory data to obtain independent serum folate, and covariate data, including total cholesterol, high-density lipoprotein, blood glucose, and glycated hemoglobin; dependent variable sleep duration and covariate data including race, family income, marital status, education, and country of birth from questionnaire data; and other covariate data such as systolic blood pressure, diastolic blood pressure, waist circumference, and body mass index data from the measured data. Data was primarily analyzed using multiple regression analysis, stratified analysis with interactions, restricted cubic splines (RCS), and threshold effect analysis. A total of 14,072 participants over 19 years of age were enrolled, comprising 6923 (49.2%) males and 7149 (50.8%) females, with a median age of 48 years. The association between serum folate levels and sleep duration exhibited an inverted L-shaped curve (P = .033) in RCS. In the threshold analysis, for participants with serum folate of <32.208 nmol/L (β = 0.608, 95% CI: (0.275, 0.942), P < .001), the sleep duration was significantly increased by 0.608 minutes, with every 1 nmol/L in serum folate increased. At serum folate of ≥32.208 nmol/L, no significant association between serum folate and sleep duration was detected [β = 5.686, 95% CI: (−1.508, 12.88), P = .118]. Our findings revealed an inverted L-shaped relationship and saturation effects between serum folate levels and sleep duration in adults in the United States. 1. Introduction Sleep is the most basic life activity of human life, and is markedly important to maintain health.[1] High-quality sleep can promote the healthy development of physiological, psychological and cognitive memory,[2] and make people happy, whereas lack of sleep or sleep disorders can lead to physical and mental illness.[2–6] Moreover, the possibility of death due to sleep deprivation has been demonstrated in animal models.[7] As early as the last century, scholars have noted the chronic lack of sleep in humans.[8] According to the American Academy of Sleep Medicine and the Society for Sleep Research, adults should get at least 7 hours of sleep to promote optimal health.[6] A study in the United States has identified a downward trend in the amount of sleep obtained by individuals, which poses a public threat to human health and longevity.[3] Folate, also known as vitamin B-9, occurs naturally in green leafy vegetables and is a necessary micronutrient that cannot synthesized by the body.[9,10] Low levels of serum folate have been documented in certain individuals. Moreover, low or deficient folate levels can lead to numerous health problems, including hematological disorders,[11] poor pregnancy outcomes,[10,12] congenital disabilities,[12,13] cognitive impairment,[12,14] and an increased stroke risk.[15] Accordingly, in 1998, the United States and Canada made it mandatory to fortify grains and flour with folate to increase the folate intake in the general populations.[14] Additionally, Beydoun et al found that lower serum folate levels can be associated with a shorter or very short sleep duration than with normal sleep duration.[16] Dilixiati et al[17] meta-analysis showed that those leading to shorter sleep duration factors may cause dysfunction of sexual function, possibly because short sleep causes impaired hypothalamic-pituitary-gonadal axis function, which causes decreased hormone level, and long-term hypoxia caused by sleep disorders such as OSA can also cause oxidative stress and neurological dysfunction in the body. As an important cofactor for carbon transfer and utilization, folate can protect telomeres, inhibit the oxidative phosphorylation process, and play an important role in nucleic acid synthesis. In addition, it can also reduce the level of cysteine and play an antioxidant effect, thus improving the disease state caused by sleep deprivation.[14] This may indirectly reflect the mechanism by which folate supplementation can improve sleep. In the current study, we used the National Health and Nutrition Examination Survey (NHANES) to assess the association between serum folate levels and sleep duration in the American adult population and to identify evidence supporting the role of folate supplements in improving sleep duration. 2. Material and methods 2.1. Study population NHANES is a program of the National Center for Health Statistics, part of the Centers for Disease Control and Prevention, that evaluates the health and nutritional status of adults and children in the United States. It is a continuous program focusing on a variety of health and nutrition measurements, and is conducted mainly in the form of interviews and questionnaires covering demographic, socio-economic, dietary, and health issues. The sample for the survey is selected to represent the U.S. population of all ages.[18] All data were collected by the National Center for Health Statistics after obtaining approval from the Research Ethics Review Board (ERB), and each participant signed an informed consent form before study commencement.[19] The current cross-sectional study utilized data from NHANES 2005 to 2010 cycles on individuals aged 19 years and older. All data are available on the NHANES website (http://www.cdc.gov/nchs/nhanes.htm). 2.2. Sleep duration and serum folate data Information on sleep duration was obtained using a household interview questionnaire. For the question item of “How many hours did you sleep?,” participants were asked to answer the question, “How much sleep do you usually get at night on weekdays or workdays?” with an integer. The number entered corresponds to the amount of time the participants slept, with the number 12 indicating that the participants slept 12 hours or more. To facilitate statistical analysis, the study converts hours into minutes. Serum folate levels (standard and SI units) were obtained from laboratory data, in which serum folate was measured using the Bio-Rad Laboratories (BRL) radiometric method in the 2005 to 2006 cycle and the microbiological method in the 2007 to 2010 cycles. In the current study, we used SI unit data and quartiles (Q1:<22.3 nmol/L, Q2: ≥22.3, and < 32.7 nmol/L, Q3: ≥32.7, and <48.6 nmol/L, Q4: ≥48.6 nmol/L) to convert it into categorical variables. To convert conventional units (ng/mL) to SI units (nmol/L), the serum value was multiplied by 2.265. 2.3. Covariates The study included basic information regarding the participants (sex and age), measured and test variables [total cholesterol, high-density lipoprotein, glucose, glycated hemoglobin, waist circumference, body mass index (BMI), and systolic and diastolic blood pressure],[5,6,20–22] and social measures (race, education, marital status, family income, and country of birth) as covariates. BMI is divided into <18.5kg/m2, 18.5kg/m2, and ≤25kg/m2, 25kg/m2, and ≤30kg/m2, and >30kg/m2.[3,23] Race was divided into Mexican Americans, other Hispanics, non-Hispanic whites, non-Hispanic blacks, and other races.[23,24] Education was divided into ≤12 grade, high school graduate, university and above. Family income was divided into ≤$35,000, $35,000 to $75,000, and ≥$75,000. Marital status was divided into married, lost spouse, and never married. Countries of birth were grouped according to their place of birth in the United States, Mexico, and other countries. 2.4. Statistical analyses The current study was a secondary analysis of public data. To maintain the data originally, missing values of all variables were removed. Categorical data are presented as percentages (unweighted), and continuous data are presented as mean ± standard deviation. In the univariate analysis, the chi-square test (categorical variables) and analysis of variance (continuous variables) were used to examine the relationship between each variable and sleep duration. A multivariate linear regression analysis was performed to examine the association between serum folate levels and sleep duration. Folate was divided into 4 groups on average, with the lowest group deemed the control group, and 3 models were included for comparison. Before inclusion in the model, we examined baseline serum folate and sleep duration. In model 1 we examined whether serum folate levels differed based on sex and age; in model 2, we continued to incorporate measure and test variables to determine whether these factors could impact serum folate and sleep duration; and finally, we included sociological factors to determine the stability of the relationship between serum folate level and sleep duration. Since the restrictive cubic spline (RCS) better captured the nonlinear relationship of the data, while serum folate was studied as a continuous variable, we observed that the 4 groups of folate in each model showed a positive direction from the lowest group to the highest group. So after obtaining the results from the above analysis, we performed restricted cubic spline curve fitting for serum folate and sleep duration with full model adjustment, identifying an inverted L-shaped relationship between them. Multiple studies[25,26] have revealed significant statistical differences between race and sleep duration in the United States, and Ghani et al reported significant differences in sleep duration and quality between individuals born in Mexico and those born in the United States.[27] Hence, we re-performed RCS curve fitting after excluding race and country of birth variables, and the relationship between serum folate and sleep duration remained an inverted L-shaped curve. Next, we performed a threshold analysis to determine the inflection point between serum folate and sleep duration. Finally, we examined the stability between serum folate and sleep duration by undertaking a subgroup analysis for sex, age, race, BMI, family income, education status, and marital status under the full model. The results revealed the presence of racial differences in sleep duration, and that non-Hispanic blacks and other racial groups were negatively associated with sleep duration, consistent with the findings of Whinnery et al[28] although other groups were stable. Data analyses were performed using the statistical software packages R. version 3.4.3 (R. Foundation for Statistical Computing, Vienna, Austria) and version 1.9.1 Free Statistics software (Beijing Free Clinical Medical Technology Co., LTD). A 2-tailed test was performed, and statistical significance was set at P < .05. 3. Results The study used data from 3 cycles (2005–2006, 2007–2008, and 2009–2010) of the NHANES. A total of 17,869 participants, aged ≥19 years, were initially included. After excluding participants with missing serum folate (n = 1751) and sleep duration (n = 7), we excluded those with missing data (n = 2039) on each covariate. Finally 14,072 participants were included in the study (Fig. 1). Flow chart of inclusion and exclusion of the study participants. Table 1 presents the baseline characteristics of 14,072 participants, comprising 6923 (49.2%) males and 7149 (50.8%) females. The group with serum folate of <22.3 nmol/L comprised a larger proportion of males, participants < 40 years, obese, non-Hispanic White, low income, high education, and born in the United States. In contrast, in the group with serum folate ≥ 32.7mmol/L, female participants had a larger proportion than male participants. This may be related to the policy that women in the US of childbearing age are given for folate supplementation.[29] It is noteworthy that non-Spanish whites were higher than any other ethnic population in any group. This should be related to the generally high social status and economic level of White Americans.[28] Table 1Characteristics of participants in the NHANES 2005 to 2010 cycles.CharacteristicParticipants*TotalQ1 (<22.3 nmol/L)Q2 (≥22.3, <32.7 nmol/L)Q3 (≥32.7, <48.6 nmol/L)Q4 (≥48.6 nmol/L) P No.14,0723491353335293519Gender Male6923 (49.2)1955 (56)1839 (52.1)1684 (47.7)1445 (41.1)<.001 Female7149 (50.8)1536 (44)1694 (47.9)1845 (52.3)2074 (58.9)Age (yr) <405100 (36.2)1620 (46.4)1473 (41.7)1270 (36)737 (20.9)<.001 ≥40, <655712 (40.6)1417 (40.6)1449 (41)1494 (42.3)1352 (38.4) ≥653260 (23.2)454 (13)611 (17.3)765 (21.7)1430 (40.6)Total cholesterol (mmol/L)5.1 ± 1.15.0 ± 1.15.1 ± 1.15.1 ± 1.15.1 ± 1.1<.001HDL(mmol/L)1.4 ± 0.41.3 ± 0.41.4 ± 0.41.4 ± 0.41.4 ± 0.4<.001Glucose (mmol/L)5.6 ± 2.15.5 ± 1.95.6 ± 2.15.6 ± 2.35.7 ± 1.9<.001Glycohemoglobin (%)5.7 ± 1.05.6 ± 1.05.7 ± 1.15.7 ± 1.05.8 ± 1.0<.001Waist (cm)98.5 ± 15.999.3 ± 17.098.8 ± 16.298.0 ± 15.398.0 ± 15.0.001BMI (kg/m2) <18.5234 (1.7)60 (1.7)58 (1.6)55 (1.6)61 (1.7)<.001 ≥18.5, <253976 (28.3)956 (27.4)958 (27.1)996 (28.2)1066 (30.3) ≥25, <304818 (34.2)1093 (31.3)1208 (34.2)1258 (35.6)1259 (35.8) ≥305044 (35.8)1382 (39.6)1309 (37.1)1220 (34.6)1133 (32.2)Systolic (mm Hg)123.2 ± 18.7122.3 ± 17.7122.8 ± 18.5122.4 ± 18.5125.3 ± 20.0<.001Diastolic (mm Hg)69.2 ± 13.370.0 ± 12.869.8 ± 13.269.1 ± 13.367.8 ± 13.7<.001Race Mexican American2691 (19.1)752 (21.5)760 (21.5)705 (20)474 (13.5)<.001 Other Hispanic1186 (8.4)220 (6.3)330 (9.3)333 (9.4)303 (8.6) Non-Hispanic White6885 (48.9)1365 (39.1)1529 (43.3)1765 (50)2226 (63.3) Non-Hispanic Black2680 (19.0)992 (28.4)753 (21.3)572 (16.2)363 (10.3) Other Race630 (4.5)162 (4.6)161 (4.6)154 (4.4)153 (4.3)Family income ($) <35,0007022 (49.9)1492 (42.7)1726 (48.9)1783 (50.5)2021 (57.4)<.001 ≥35,000, <75,0004142 (29.4)931 (26.7)926 (26.2)1106 (31.3)1179 (33.5) ≥75,0002908 (20.7)1068 (30.6)881 (24.9)640 (18.1)319 (9.1)Educations Less (include) 12th Grade4032 (28.7)845 (24.2)959 (27.1)1020 (28.9)1208 (34.3)<.001 High School3505 (24.9)687 (19.7)840 (23.8)914 (25.9)1064 (30.2) College or above6535 (46.4)1959 (56.1)1734 (49.1)1595 (45.2)1247 (35.4)Marital status Married9107 (64.8)2013 (57.7)2224 (63)2325 (65.9)2545 (72.3)<.001 Lose Spouse3138 (22.3)835 (24)763 (21.6)777 (22)763 (21.7) Never married1814 (12.9)638 (18.3)541 (15.3)425 (12)210 (6)Born country Born in US States10,627 (75.5)2589 (74.2)2564 (72.6)2655 (75.2)2819 (80.1)<.001 Born in Mexico1653 (11.7)522 (15)474 (13.4)409 (11.6)248 (7) Born in Other Country1792 (12.7)380 (10.9)495 (14)465 (13.2)452 (12.8)Abbreviations: BMI = body mass index, HDL = high-density lipoprotein, Q1 to Q4 = quartile based on serum folate.*Data are presented as unweighted number (weighted percentage) for categorical variables and mean (SE) for continuous variables. Characteristics of participants in the NHANES 2005 to 2010 cycles. Abbreviations: BMI = body mass index, HDL = high-density lipoprotein, Q1 to Q4 = quartile based on serum folate. Data are presented as unweighted number (weighted percentage) for categorical variables and mean (SE) for continuous variables. Table S1, Supplemental Digital Content, http://links.lww.com/MD/O87, shows the relationship between covariates and sleep duration, revealing that sex, age, high-density lipoprotein, glycated hemoglobin, BMI, diastolic blood pressure, race, and country of birth were related. However, education, family income and marital status showed no significant correlation, which is in contrast to the findings of Luo et al,[1] who found significant differences in education and sleep duration between White and Black adults. After adjusting for all confounders, a positive correlation was observed between serum folate levels and sleep duration (Table 2). Compared with the lowest quartile of the serum folate group Q1 (<22.3 nmol/L), the respectively β values of Q2 (≥22.3 and <32.7 nmol/L), Q3 (≥32.7 and <48.6 nmol/L), and Q4 (≥48.6 nmol/L) were 4.4 (95% CI: 0.42, 8.39), 6.29 (95% CI: 2.24, 10.35), and 7.92 (95% CI: 3.63, 12.22). Not included the variables of race and country of birth, the fit curve of the RCS model (Figure S1, Supplemental Digital Content, http://links.lww.com/MD/O88) (P = .002) showed an inverse L-shaped relationship between serum folate levels and sleep duration(P = .033). The RCS fitting curves in the full model (Fig. 2) showed that this curve relationship persisted. Based on the threshold analysis, in participants with serum folate levels of <32.208 nmol/L [β = 0.608, 95% CI: (0.275, 0.942), P < .001], the sleep duration was significantly increased by 0.608 minutes with every 1 nmol/L increase in serum folate. However no significant association between serum folate and sleep duration was observed when the serum folate was ≥32.208 nmol/L [β = 5.686, 95% CI: (−1.508, 12.88), P = .118] (Table 3). This may provide a new idea for clinicians in treating sleep problems, especially in patients with short sleep duration. Serum folate as a routine test item and with 32.208 nmol/L as a cutoff value, and be treated with folate supplementation for those patients severely below this threshold, thereby improving sleep. Of course, this requires further exploration through rigorous longitudinal research. Table 2Association between serum folate and sleep duration among participants in the NHANES 2005 to 2010 cycles.No.crudeP-valueModel 1P-valueModel 2P-valueModel 3P-valueβ (95% CI) Serum folate140720.2 (0.15–0.25)<.0010.15 (0.09–0.2)<.0010.13 (0.08–0.19)<.0010.1 (0.04–0.15)<.001Quartile Q134910 (Ref).0020 (Ref).0040 (Ref).0040 (Ref).03 Q235336.38 (2.35–10.4)5.96 (1.96–9.97)5.94 (1.94–9.94)4.4 (0.42–8.39) Q335299.73 (5.71–13.75)<.0019.02 (4.99–13.04)<.0018.6 (4.58–12.63)<.0016.29 (2.24–10.35).002 Q4351915.04 (11.02–19.07)<.00112.15 (7.98–16.31)<.00111.33 (7.16–15.5)<.0017.92 (3.63–12.22)<.001Note: Model 1 was adjusted for age, gender.Model 2 was adjusted for model 1 + total cholesterol + HDL + glucose + glycohemoglobin + waist + BMI + systolic + diastolic.Model 3 was adjusted for model 2 + race + family income + education + marital status + born country.Abbreviations: β = standardized regression coefficients, CI = confidence interval, HDL = high-density lipoprotein, Q = quartile. Association between serum folate and sleep duration among participants in the NHANES 2005 to 2010 cycles. Note: Model 1 was adjusted for age, gender. Model 2 was adjusted for model 1 + total cholesterol + HDL + glucose + glycohemoglobin + waist + BMI + systolic + diastolic. Model 3 was adjusted for model 2 + race + family income + education + marital status + born country. Abbreviations: β = standardized regression coefficients, CI = confidence interval, HDL = high-density lipoprotein, Q = quartile. Table 3Threshold effect analysis of the relationship of serum folate with sleep duration.Serum folate levels (nmol/L)Adjusted ModelThreshold of serum folateβ (95% CI)P-value<32.2080.608 (0.275, 0.942)<.001≥32.2085.686 (−1.508, 12.88).118Likelihood ratio test.002Adjusted for age, gender, total Cholesterol, HDL, glucose, glycohemoglobin, waist, BMI, systolic, diastolic, race, family income, education, marital status, borncountry.100% data were shown.Abbreviations: β = Standardized regression coefficients, CI = confidence interval, HDL = high-density lipoprotein. Threshold effect analysis of the relationship of serum folate with sleep duration. Adjusted for age, gender, total Cholesterol, HDL, glucose, glycohemoglobin, waist, BMI, systolic, diastolic, race, family income, education, marital status, borncountry.100% data were shown. Abbreviations: β = Standardized regression coefficients, CI = confidence interval, HDL = high-density lipoprotein. Association between serum folate and sleep duration. Solid and dashed lines represent the predicted value and 95% confidence intervals. All covariates were included in the adjusted model. Only 99% of the data is shown. Moreover, we performed stratified analyses in several subgroups to assess the potential impact of changes in the relationship between serum folate levels and sleep duration. No significant interaction was found after stratification by sex, age, BMI, education, family income, and marital status. However, race could impact sleep duration, particularly in non-Hispanic blacks and other ethnic groups, similar to that reported by Sheehan et al[26] (Fig. 3). Stratified multivariable analysis of the association between serum folate and sleep duration according to gender, Age, BMI, family income, education, marital status, race. BMI = body mass index. 4. Discussion Herein, we identified a positive correlation between serum folate levels and sleep duration. Given that sleep duration is as a common measure of health-related problems, the American Academy of Sleep Medicine recommends that adults get between 7 and 9 hours of sleep daily.[6] However, reduced sleep duration has become a common social phenomenon,[30] and is closely associated with health risks. Studies found that lack of adequate sleep can alter an individual’s food intake and preferences,[31] decreases leptin, increases ghrelin, and decreases insulin sensitivity,[19] and has also been implicated in obesity among preschoolers.[32] Prolonged low levels of serum folate also pose health risks. In particular, neural tube defects and adverse pregnancy outcomes. Beydoun et al[16] found that low levels of serum folate are associated with restless leg syndrome in pregnant women. Vrieling et al[33] even reported that low levels of serum folate are associated with an increased risk of urothelial cell carcinoma. Folate, as a micronutrient solely derived from extemal sources, is primarily associated with neural tube defects in early development.[29] In 1991, the Vitamin Research Group identified the specific role of folate in preventing neural tube defects and recommended public measures to ensure that women of childbearing age receive adequate folate supplements.[29] This explains our findings why serum folate levels are generally higher in women than in men. Subsequently, Ren et al[34] identified an inverse association between increased folate intake and the risk of breast cancer. Accordingly, would folate supplementation improve sleep quality in individuals with low folate levels? Our results suggest that supplementation is a feasible strategy, although a saturation effect was observed. At a serum folate level of ≥32.208 nmol/L, continued supplementation may fail to improve sleep, with high folate levels posing potential health risks. For example, high folate levels in pregnant women may increase the risk of obesity in large-gestational-age babies and male offspring.[35] A prospective study in the United States[36] has revealed that a high folate status may promote the development of colorectal cancer, although this effect was dependent on the dose and time of exposure. Smith et al[37] found that folate is a double-edged sword that can prevent cancer while potentially promoting the growth of pre-tumor cells and the progression of subclinical cancers. In a meta-analysis,[38] the authors found that sexual dysfunction was associated with prostate cancer, and that individuals with sexual dysfunction were more likely to have prostate cancer. There are many factors that cause sexual dysfunction, including various sleep problems such as poor sleep quality and sleep disorders. The authors[39] also found that sleep problems disrupt the physiological processes that regulate the immune system, which in turn can excessively increase inflammatory responses and accelerate disease progression. Another study[16] found that the control center of the sleep wake cycle was located in the suprachiasmatic nucleus, and folate methylates homocysteine to methionine, resulting in nucleic acid methylation, which correlated with the circadian rhythm of clock proteins and neuropeptide vasopressin located in the suprachiasmatic nucleus. On the other hand, it needs to be further explored whether people with low levels of serum folate can improve sleep quality through folate supplementation, which can block or delay the process of related cancer. Therefore, our results may provide a reference for the use of folate supplements in individuals with serum folate deficiency. Nevertheless, the limitations of our study need to be addressed. First, as a cross-sectional study, our data was derived from the US public database NHANES 2005 to 2010 cycles. Owing to the mandatory addition of folate to cereal products and flour in the United States and Canada in 1998, levels of folate in the current population have generally increased, which may lead to differences between our results and those of the current population. Second, in screening for covariates, we neglected the population behavior variable, such as smoking,[3,40] drinking,[3,41] and drug use.[42] In fact, population behavior is an important class of variables, however, given the considerable data loss upon extraction, we excluded this class of variables in the current study to ensure our sample size and avoid data interpolation bias. We plan to address this constraint in future studies to obtain more valuable results. Finally, compared to the sleep duration detected by the activity recorder, the self-reported sleep duration may be erroneous although some studies have shown good agreement between the activity recorder and self-reported duration.[20] Despite these shortcomings, we employed a sufficiently large sample size and the earliest data set. To the best of our knowledge, is the first study to quantitatively analyze the relationship between serum folate levels and sleep duration, however, prospective studies are needed to confirm this relationship. 5. Conclusion The findings of the current study highlight the positive correlation between serum folate levels and sleep duration in the American adult population, although a saturation effect was observed. This quantitative result suggests that in countries with mandatory folate supplementation, serum folate levels need residents be checked regularly to avoid the harm caused by high levels of folate. At the same time, for people suffering from sleep problems, serum folate tests can be included in routine item to assess folate levels and provide targeted folate supplements. Of course, further studies are needed to confirm the identified association. Acknowledgments We appreciate the support of Dr Jie Liu of the Department of Vascular, Chinese PLA General Hospital, and Dr Huanxian Liu of the Department of Neurology, the First Medical Center, Chinese PLA General, Beijing, China, for assistance with the statistical and study design consultations. Author contributions Conceptualization: Yandong Dong. Data curation: Yandong Dong, Mei Luo. Methodology: Yandong Dong, Mei Luo. Validation: Yandong Dong. Writing – original draft: Yandong Dong. Writing – review & editing: Yandong Dong, Mei Luo. Supplementary Material