. 2016 Oct 5;20(4):687–701. doi: 10.1017/S1368980016002603

Table 1.

Observational studies on micronutrients and sleep patterns in human subjects

Study	Methods	Samples/settings	Micronutrient measures	Sleep measures	Major findings	Strengths	Limitations
Luojus et al. (2015)⁽ ⁴³ ⁾	Cross-sectional study V: age, cumulative smoking history, alcohol consumption, Human Population Laboratory depression scale scores, physical activity, cardiometabolic syndrome, CVD history	Men aged 42–60 years (n 2570)/Eastern Finland	Serum Zn and Cu concentrations	Self-reported sleep duration	Sleep duration significantly associated with level of both serum Cu and hs-CRP in adjusted models. Zn no longer significantly associated with sleep duration after adjustment	1. A large sample size 2. Biomarkers of Zn and Cu 3. Controlled for important confounders	1. Self-reported sleep duration may lead to recall bias 2. Sleep duration was categorized into nine levels, which decreased statistical power 3. Only included men in the study, which may reduce its generalizability
Ji & Liu (2015)⁽ ⁴² ⁾	Longitudinal study V: grade, sex, education level of mother and father	Followed children from 3–5 years to 11–15 years old (n 1295)/China	Serum Zn concentrations	Self-reported sleep patterns, Pittsburgh Sleep Quality Index	Cross-sectional analyses showed negative correlations of blood Zn concentrations with insufficient sleep duration, sleep disturbances and poor sleep quality in adolescence, but no association at pre-school age. Longitudinal analyses indicated that blood Zn concentrations at pre-school age predict poor sleep efficiency and poor sleep quality in adolescence	1. Includes two-wave longitudinal data from a pre-school cohort 2. Large sample size 3. Used biomarker of Zn status	1. Used only subjective sleep measures 2. Did not adjust for dietary nutrients, environmental influential factors
Massa et al. (2015)⁽ ⁵⁵ ⁾	Cross-sectional study V: age, clinic, season, comorbidities, BMI, physical and cognitive function	Men aged 68 years or older (n 3048)/USA	Total 25(OH)D combining 25(OH)D₂ and 25(OH)D₃	Nightly total sleep time, sleep efficiency and WASO obtained using wrist actigraphy worn for an average of five consecutive 24 h periods	Low levels of total serum 25(OH)D are associated with poorer sleep including short sleep duration and lower sleep efficiency	1. Objective sleep measures for an average of five consecutive 24 h periods 2. Used biomarkers of nutrient 3. Data were from a large cohort study	1. Findings may not be generalizable to young men or women
Bertisch et al. (2014)⁽ ²⁰ ⁾	Cross-sectional study V: Model A adjusted for age, sex, race/ethnicity, examination site, waist circumference; Model B additionally adjusted for education, family income, physical activity, smoking status, alcohol intake; Model C additionally adjusted for antidepressant use, depression score, history of osteoarthritis, history of asthma, glomerular filtration rate	Adults with mean age of 68·2 years (n 1721)/USA	Serum 25(OH)D concentration	Sleep duration, efficiency and symptoms were measured by PSG, actigraphy and questionnaires	Vitamin D-deficient individuals slept shorter than sufficient individuals, with strongest associations shown in African Americans. Chinese Americans with vitamin D deficiency had higher AHI v. sufficient individuals	1. Investigated ethnic disparity in sleep effect of vitamin D 2. Used both objective and subjective sleep measures 3. Used biomarker of nutrient 4. Rigorously controlled for multiple potential confounders	1. Vitamin D measurement preceded the collection of sleep outcomes by average of 10·3 years 2. Did not adjust for season variation of vitamin D levels
Kim et al. (2014)⁽ ⁵⁴ ⁾	Cross-sectional study V: sex, age, cigarette smoking, alcohol consumption, self-reported daily sun exposure	Adults aged 60–80 years (n 1614)/South Korea	Serum 25(OH)D concentration	Self-reported sleep duration	Serum vitamin D level positively associated with self-reported daily sleep duration in elderly Korean individuals	1. Used a nationally representative sample of older adults 2. Considered sun exposure as a vitamin D resource	1. Did not adjusted for dietary intake, supplementation of vitamin D and season influence 2. Self-reported sleep duration and sun exposure may lead to recall bias
Beydoun et al. (2014)⁽ ⁵² ⁾	Cross-sectional study V: age, sex, race/ethnicity, education, marital status, family income, BMI, smoking, physical activity, self-reported chronic conditions, antidepressant medication use, dietary intakes	US NHANES, 2005–2006 (aged 20–85 years, n 2459)/USA	Serum concentrations of key nutrients	Sleep questionnaire included items on sleep habits and disorders; and a subscale of the Functional Outcomes of Sleep Questionnaire	Independent inverse associations were found between serum vitamin B₁₂ and sleep duration, between 25(OH)D and sleepiness (as well as insomnia), and between folate and sleep disturbance	1. Used a nationally representative sample 2. Biomarkers of nutrients 3. Adjusted for important confounders	1. Did not stratify age groups 2. Self-reported sleep may lead to recall bias
Shiue (2013)⁽ ⁵³ ⁾	Cross-sectional study V: sex, ethnicity, BMI, high blood pressure, active smoking, depressive symptoms	NHANES, 2005–2006 (aged 16 years or above, n 6139)/USA	Serum 25(OH)D concentrations	Self-reported sleeping hours, minutes to fall asleep and sleep complaints; sleep disorders diagnosed by doctors	No association between serum 25(OH)D concentrations and sleeping hours was observed, while a significant inverse association was found between serum 25(OH)D concentrations and minutes to fall asleep. Moreover, people with higher vitamin D levels could be more likely to complain about sleep problems, although the reason is unclear	1. Used a nationally representative sample	1. Did not stratify by age group 2. Did not adjust the status of other nutrients which may affect sleep patterns 3. Self-reported sleep duration and latency may lead to recall bias

Grandner et al. (2013)⁽ ³⁵ ⁾	Cross-sectional study V: total energy intake, total number of foods consumed, age, gender, income, education, BMI, exercise	Adults aged 18+years in NHANES 2007–2008 (n 5587)/USA	24 h food recall	Self-reported sleep duration	After adjustment for overall diet, only decreased P, Mg, Fe, Zn and Se in the context of very short (<5 h) sleep and decreased P in the context of long sleep remained significant. Some of the effects of the vitamin for very short (<5 h) sleep remained including decreased thiamin, total folate, folic acid and folate DFE in fully adjusted analyses	1. Sampling was performed to ensure generalizability to the US population 2. Effects of covariates were examined separately and totally	1. Self-reported sleep duration and dietary intake may lead to recall bias 2. Did not differentiate sleep duration on weekdays and weekends

Song et al. (2012)⁽ ⁴¹ ⁾	Cross-sectional study V: age, smoking status, occupation	Women aged 21–72 years (n 126)/South Korea	Zn, Cu and Zn:Cu in serum and hair	Self-reported questionnaire (7 d recall): average sleep hours separately on weekdays and weekends	Participants in the middle tertile of Zn and Zn:Cu ratio in serum had significantly longer sleep duration v. those in the lowest tertile. Increasing Zn:Cu in hair was associated with longer sleep hours, whereas sleep duration decreased significantly from the lowest to the highest tertile of hair Cu level	1. Compared results from different micronutrient measures	1. Results of the study may not generalize to men and other people 2. Self-reported sleep duration may lead to recall bias 3. Did not mention the method used to calculate mean hours of sleep based on weekday and weekend sleep hours

Sato-Mito et al. (2011)⁽ ⁴⁸ ⁾	Cross-sectional study V: current smoking status, supplement use, energy intake	Women aged 19–36 years (n 112)/Japan	Self-administered diet history questionnaire (1-month recall)	MEQ, preferred bedtime and rise time	Lower MEQ score showed significant associations with lower energy-adjusted intakes of protein, Ca, Mg, Zn, vitamins (D, riboflavin and B₆) and vegetables, and with a higher intake of noodles. Furthermore, later midpoint of sleep showed significant associations with lower energy-adjusted intakes of protein, cholesterol, K, Ca, Mg, Zn, vitamins (D, riboflavin, B₆ and B₁₂), soya, fish and shellfish, and eggs, and with higher intakes of noodles, bread and confections	1. Supplement use was addressed in data analyses	1. Sample used may be limited in its generalizability 2. Self-reported dietary intake may lead to recall bias 3. Investigated preferred rather than actual sleep schedule 4. Did not consider known confounders such as SES, course load
Sato-Mito et al. (2011)⁽ ⁴⁰ ⁾	Cross-sectional study V: means of dietary intake, dietary behaviours (time at which meals began, eating duration, number of skipped meals, number of occasions when TV was watched during weekday meals), lifestyle variables, residential block, size of residential area and current smoking status	Female dietetics students aged 18–20 years (n 3304)/Japan	Self-administered diet history questionnaire (1-month recall)	Self-reported bedtimes and rise times on weekdays	Late midpoint of sleep was significantly negatively associated with the percentage of energy from protein and carbohydrates, and energy-adjusted intakes of cholesterol, K, Ca, Mg, Fe, Zn, vitamin A, vitamin D, thiamin, riboflavin, vitamin B₆, folate, rice, vegetables, pulses, eggs, and milk and milk products	1. Used large sample size 2. Important confounders were controlled for	1. Sample used may be limited in its generalizability 2. Self-reported dietary intake may lead to recall bias 3. Did not consider sleep-latency-adjusted midpoint of sleep 4. Did not consider known confounders such as SES, course load 5. Investigated sleep schedule on weekdays only
Grandner et al. (2010)⁽ ⁵⁶ ⁾	Cross-sectional study V: age, income, education, total dietary amount, BMI, minutes of moderate–strenuous physical activity	Postmenopausal women (n 459)/USA	FFQ (3-month recall)	Objective sleep: 1 week of actigraphy. Subjective sleep: sleep diary	Later sleep acrophase, an indicator of sleep timing, was associated with more dietary vitamin D	1. Used large sample size 2. Incorporated subjective and objective sleep measurement	1. Sample used may be limited in its generalizability 2. Self-reported dietary intake may lead to recall bias 3. Did not differentiate weekdays from weekends for sleep schedule
Kordas et al. (2008)⁽ ³⁴ ⁾	Cross-sectional study V: age, sex, SES, breast-feeding, caste, malarial parasite count, Trichuris egg densities, illness in previous week, walking unassisted	Infants aged 6–18 months: Study 1 (n 174)/Pemban and Zanzibar Study 2 (n 770)/Pemban and Zanzibar Study 3 (n 326)/Nepal	Hb, serum Zn protoporphyrin	Sleep questionnaire (parental report)	IDA infants were associated with shorter night-time sleep duration and higher frequency of night waking	1. Used large sample size. 2. Used biomarkers of micronutrient status 3. Compared results across three places	1. Parental reports/recall may lead to measurement bias 2. Did not mention if infants with medication were excluded 3. Adjusted for different covariates in the studies, which inhibited the precision of the comparison across studies
Kordas et al. (2007)⁽ ³⁹ ⁾	Cross-sectional study V: anxiety, bedroom sharing, who decides when to get up and go to bed, age, sex, crowding in the home	Children aged 6–8 years (n 550)/Mexico	Serum Zn, SF, Hb	Sleep questionnaire (parental report)	Children with anaemia tended to have an earlier bedtime and be less likely to have long sleep-onset latency. Low SF was related to longer sleep-onset latency. Zn deficiency was not related to sleep, behaviour or activity	1. Used large sample size 2. Used biomarkers of micronutrient status	1. Parental recall may lead to measurement bias 2. Did not differentiate weekdays from weekends for sleep schedule 3. Did not classify the type of anaemia
Peirano et al. (2007)⁽ ³⁶ ⁾	Cross-sectional study V: age, gender, birth weight, weight-for-age Z-score, mother’s IQ	Otherwise healthy 6-month-old infants with IDA (n 26) and non-anaemic control infants (n 18)/Chile	Hb, SF	Sleep PSG (EEG activity)	IDA infants differed from control group by having sleep spindles with reduced density, lower frequency and longer inter-spindle intervals in NREM sleep stage 2 and SWS	1. Used objective sleep measurement	1. Did not evaluate the precise location of spindle waves which reflect different developmental paths 2. Did not tease apart the pharmacological effects
Peirano et al. (2007)⁽ ³⁷ ⁾	Longitudinal cohort study Exposure: IDA in infancy V: age, sex	Healthy 4-year-old children (n 55; former IDA, n 27, non-anaemic controls, n 28)/Chile	IDA in infancy	Sleep PSG (EEG activity)	Relative to controls, former IDA children showed: (i) longer duration of REM sleep episodes in the first third and shorter in the last third; (ii) more REM sleep episodes in the first third and fewer in the second third; and (iii) shorter latency to the first REM sleep episode and shorter NREM stage 2 and SWS episodes within the first sleep cycle	1. Time–order relationship was generally clear 2. Used objective sleep measurement	1. Only recorded a single night’s sleep in the laboratory 2. Did not assess daytime naps and could not determine whether disrupted night-time sleep was caused by a long daytime nap 3. Did not consider current Fe status
Lichstein et al. (2007)⁽ ⁵⁷ ⁾	Cross-sectional study V: age, ethnicity, sex	People ranging in age from 20 to 98 years (n 72)/USA	Self-reported vitamin use (1-month recall)	Sleep diaries and sleep questionnaires	For those individuals taking a multivitamin or multiple single vitamins, sleep diaries revealed poorer sleep v. non-vitamin users in the number and duration of awakenings during the night. After controlling for age, ethnicity and sex, the difference in the number of awakenings was still marginally significant	1. Used large sample size	1. Self-reported vitamin intake may lead to recall bias 2. Sleep diary may be not reliable 3. Did not stratify the type of vitamin use 4. Did not collect data on herbal supplements
Black et al. (2006)⁽ ⁴⁵ ⁾	Longitudinal cohort study Exposure: antenatal MgSO₄ supplement V: N/A	Preterm infants (n 134): MgSO₄ only (n 5) Steroids only (n 46) MgSO₄ + steroids (n 45) Non-treatment (n 38)/USA	MgSO₄ supplement	Sleep EEG	Infants exposed to MgSO₄ had more active sleep without REM. MgSO₄-only group had higher quiet sleep regularity scores and fewer state changes	1. Time–order relationship was generally clear 2. Used objective sleep measurement	1. Did not mention if confounders were controlled for in regression models 2. Dose or duration of MgSO₄ treatment was not adjusted for 3. Sampling issues may limit the generalizability of these findings 4. Size of the MgSO₄ only group is small
Dralle & Bodeker (1980)⁽ ⁴⁴ ⁾	Cross-sectional studyV: N/A	Full-term newborn infants between the 5th and 15th days of life (n 14)/Germany	Serum Mg	Sleep EEG, EMG from chin muscles, EOG and ECG	With increasing serum Mg, quiet sleep increased, whereas active sleep decreased	1. Used objective sleep and micronutrient measures	1. Used small sample size 2. Mg treatment was not controlled for in regression models

V, variables adjusted for; TV, television; SES, socio-economic status; IQ, intelligence quotient; IDA, Fe-deficiency anaemia; N/A, not applicable; NHANES, National Health and Nutrition Examination Survey; 25(OH)D, 25-hydroxyvitamin D; 25(OH)D₂, 25-hydroxyergocalciferol; 25(OH)D₃, 25-hydroxycholecalciferol; SF, serum ferritin; WASO, wake time after sleep onset; PSG, polysomnography; MEQ, Morningness–Eveningness Questionnaire; EEG, electroencephalography; EOG, electro-oculogram; ECG, electrocardiography; hs-CRP, high-sensitivity C-reactive protein; AHI, apnaea–hypopnoea index; DFE, dietary folate equivalents; NREM, non-rapid eye movement; SWS, slow-wave sleep; REM, rapid eye movement.