SUMMARY
Background:
Use of melatonin supplements has been increasing substantially in both children and adults in the US; however, their long-term cardiometabolic effects remain unclear. We aimed to assess the associations between regular use of melatonin supplements and the risk of developing type 2 diabetes mellitus (T2DM) or cardiovascular disease (CVD) in adults.
Methods:
Prospective cohort study of 67,202 women from the Nurses’ Health Study (NHS, 1998–2021, age at baseline: 63.6±7.1 years), 65,241 women from the NHS II (2003–2023, 48.2±4.7 years) and 26,629 men from the Health Professionals Follow-up Study (HPFS, 1998–2020, 62.9±8.8 years). Regular use of melatonin supplements was self-reported. Outcomes included CVD and T2DM incidence.
Findings:
Melatonin supplement use in the study cohorts doubled over recent decades from <2% in 1998–2007 to ≥4% in 2014–2015 (4.0% in men and 5.3% in women). During 2-million-person years of follow-up, we documented a total of 16,917 incident CVD events and 12,730 incident cases of T2DM. In a pooled analysis of the three cohorts, comparing the users with non-users of melatonin supplements, the pooled multivariable-adjusted hazard ratios (HRs) were 0.94 (95% CI, 0.83–1.06, P=0.32) for CVD and 0.98 (95%CI: 0.86–1.12, P=0.80) for T2DM. In secondary analyses, melatonin supplement use appeared to attenuate the positive association between long-term shift work (5+ years) and risk of CVD (Pinteraction=0.013) among the nurses.
Interpretation:
With up to 23 years of follow-up of 3 large prospective cohorts of middle-aged men and women, self-reported melatonin supplement use was not associated with the risk of T2DM or CVD. Further research is warranted to assess if melatonin supplement use could mitigate the potential risks of T2DM and CVD associated with rotating night shift work.
Funding:
NIH grants UM1 CA186107, U01 CA176726, U01 CA167552, U01 HL145386, R01 HL034594, R01 HL088521, and R01 HL35464
Introduction
Melatonin is a hormone secreted mainly by the pineal gland and an output signal of the circadian clock, which if exogenously applied, can help synchronize the body’s daily circadian rhythms with the light-dark cycle.1 Oral melatonin supplements, most typically in dosages of 3 or 5 mgs, are widely available in the US, either by prescription or simply over the counter. Melatonin is primarily advertised as a natural sleep aid for general use, but also for improving sleeping difficulties due to circadian rhythm disturbances in night shift workers, or for easing jet lag.1,2 Based on data from the US National Health and Nutrition Examination Surveys (NHANES), in the last two decades, use of melatonin supplements has significantly increased from 0.4% to 2.1% among adults3 and from 0.1% to 0.7% among children and adolescents in the US,4 corresponding to more than six million Americans using melatonin supplements.
The growing use of exogenous melatonin in the general population prompts a careful consideration of potential negative side effects.1,2 Contrary to the findings from observational studies supporting that higher nocturnal melatonin secretion is associated with lower blood pressure,5 or lower risk of insulin resistance,6 diabetes7 and myocardial infarction,8 several small studies (n<=21) found that controlled immediate-release melatonin administration was associated with impaired glucose tolerance and worsening insulin sensitivity,9–11 although prolonged-release melanin was found to be beneficial in a similar setting.12 Melatonin use has also been associated with increased blood pressure during the day in patients concurrently taking melatonin supplements and antihypertensive medication, while a nocturnal decrease of blood pressure was simultaneously observed in this patient group of non-dippers with coronary artery disease.13 Most short-term studies have found positive cardiometabolic effects.14,15 Specifically, in a meta-analysis of 16 clinical trials ranging from 4 to 24 weeks of supplementation with melatonin, nine studies showed benefits in diabetes-related biomarkers compared with placebo, including lowering glucose and insulin resistance.14 Overall, although the current evidence for short-term cardiometabolic effects of melatonin supplements is relatively consistent, evidence on long-term risk of chronic diseases is lacking.
In this study, we examined the association of melatonin supplement use with risk of type 2 diabetes (T2DM) and cardiovascular disease (CVD) by using repeated measurements of melatonin supplement use, detailed lifestyle factors, sleep related indicators and other potential confounders during up to 23 years of follow-up. Our analyses included men and women from three large ongoing prospective US cohorts: the Nurses’ Health Study (NHS), the Health Professionals’ Follow-up Study (HPFS) and the NHSII. We also conducted secondary analyses stratified by duration of rotating night shift work in NHS and NHS II, to examine whether the associations with melatonin supplement use differed by rotating night shift work.
Methods
Study population
The NHS,16 HPFS16 and NHS II17 were established in 1976, 1986, and 1989, respectively (appendix). Detailed information on lifestyle habits and medical history was updated biennially. Participants in NHS, NHS II and HPFS completed an initial question about regular melatonin supplement use (yes/no) in 1998, 2003, and1998, respectively, which served as the baseline for this analysis. Melatonin supplement use was updated approximately every four years thereafter in NHS and NHSII and every two years in HPFS. The analysis reported here included participants who were free of cancers at baseline and responded to the long version questionnaires which included the question about use of melatonin supplement. We further excluded baseline prevalent CVD for incident CVD analysis and baseline prevalent diabetes for incident diabetes analysis, which left a sample size of 159,072 for CVD and 157,149 for T2DM analyses.
The study protocol was approved by the Institutional Review Boards of the Brigham and Women’s Hospital and the Harvard T.H. Chan School of Public Health, and the return of completed questionnaires was considered implied consent.
Assessment of melatonin supplementation and sleep-related factors
In the section of the questionnaire on supplement use, participants were asked to choose “are there other supplements that you take on a regular basis?” followed by a list of supplements including “melatonin”, beginning in 1998 and repeated in 2002, 2006 and 2010 for NHS, in 2003, 2007, 2011 and 2015 for NHSII, and in 1998, 2000, 2002, 2004, 2006, 2008, 2010 and 2014 for HPFS. The information on brand or dosage of melatonin supplements was not collected. We stopped collecting data on melatonin supplement use after 2014/2015 in the cohorts.
Frequency of difficulty falling asleep or staying asleep was reported in 2000 in NHS, 2001 and 2013 in NHSII, and 2004 and 2012 in HPFS. The number of years having worked rotating night shifts (at least three nights per month in addition to days/evenings in that month) was only recorded for nurses (1988 in NHS, and 1989, 1991, 1993, 2001, 2005, 2007, 2009, 2011 and 2013 in NHSII).
Assessment of other related factors
We used the main biennial follow-up questionnaires to collect and update information on age, family history of myocardial infarction and diabetes, body weight, cigarette smoking, physical activity, use of aspirin, multivitamin use, coenzyme Q10 and fish oil supplement use, sleep duration, frequency of snoring, and history of hypercholesterolemia, hypertension, and depression. We considered participants to have hypercholesterolemia, hypertension, or depression when they reported these conditions on the biennial questionnaire or when they reported use of lipid or blood pressure-lowering drugs or antidepressant medications. Height, race, and ethnicity were assessed once in each cohort. We also gathered information on menopausal status and use of postmenopausal hormones for women. We assessed alcohol intake and food intakes through validated food frequency questionnaires. Diet quality was assessed by the Alternate Healthy Eating Index (AHEI-2010).16 Information of occupational stress and control was collected with Karasek and Theorell’s job content questionnaire, a self-administered 27-item instrument to assess psychological workload (demand) of a job and level of control available for managing the workload.18
Assessment of incident type 2 diabetes
Every 2 years, participants in all three cohorts were asked whether they had any physician-diagnosis of diabetes. Participants who reported physician-diagnosed diabetes were sent a supplementary questionnaire to obtain information about symptoms, diagnostic tests, and diabetes drug use to confirm a diagnosis of diabetes and the date of diagnosis.
The diagnosis was confirmed if at least one of the following was reported according to the National Diabetes Data Group criteria:19 1) at least one symptom (excessive thirst, polyuria, weight loss, or hunger) plus fasting glucose concentration of ≥7.8 mmol/L or random glucose concentration of ≥11.1 mmol/L; 2) in the absence of symptoms, at least two elevated glucose concentrations on different occasions (fasting glucose concentration >7.8 mmol/L, random glucose concentration ≥11.1 mmol/L, and/or 2-h post load concentration ≥11.1 mmol/L at an oral glucose tolerance test); or 3) treatment with insulin or oral hypoglycemic medication. After June 1998, the diagnostic criteria of type 2 diabetes were changed and a fasting blood glucose level of 7 mmol/l, rather than 7.8 mmol/l, was used as the threshold for diagnosis19 (appendix).
Assessment of incident cardiovascular disease
CVD outcomes included fatal and non-fatal coronary heart disease (CHD) and coronary revascularization (coronary artery bypass graft surgery and percutaneous coronary intervention), as well as fatal and nonfatal stroke.22 When a participant reported an incident event on any of the biennial questionnaires, permission was requested to examine medical records, which were reviewed by study investigators blinded to the participant’s risk factor status (appendix).
Statistical Analysis
Age-standardized proportion of melatonin supplement users was presented in each survey year of each cohort. Direct standardization was used to describe the proportion of melatonin supplement users in each survey cycle using the age distribution of whole cohort.
For CVD outcomes, we calculated person-years from the date of return of the baseline questionnaire (The starting time among individual participants varied within 2-year range of the study cycle) to the date of diagnosis of CVD, death, or the end of the follow-up (June 30, 2019, for the NHS; June 30, 2019, for the NHS II, and January 31, 2020, for the HPFS), whichever came first. For incident T2DM, we calculated person-years from the date of return of the baseline questionnaire to the date of diagnosis of T2DM, death, last response to the survey, or the end of the follow-up (June 30, 2021, for the NHS; January 31, 2023, for the NHS II, and January 31, 2020, for the HPFS), whichever came first (appendix).
We used time-dependent Cox proportional hazards regression to estimate the hazard ratio (HR) for regular users of melatonin supplements versus non-users in relation to risk of CVD and T2DM separately, with an origin time of the return date of the 1998 questionnaire for NHS and HPFS and return date of the 2003 questionnaire for NHS II. The basic model included age, calendar time with updated information at each 2-year questionnaire cycle, and cohorts (only for pooled analysis). The second model adjusted for race (Caucasian or not), ethnicity (Hispanic or not), use of aspirin, use of multivitamins, coenzyme Q10, or fish oil supplements, family history of diabetes or MI, and menopausal status and estrogen hormone therapy (no use, past use, current use or missing, women only), smoking status (never, ever, current smoking, or missing), alcohol intake (0, 0.1–14.9, or ≥15 g/day of alcohol), AHEI (tertile) and physical activity levels (<3, 3–8, 9–17, 18–26, 27–41, and 42+ MET-hours/week), baseline body mass index (BMI) of each 2 years cycle (<18.5, 18.5–22.9, 23–24.9, 25–29.9, 30–34.9, 35+ kg/m2) and weight changes within each 2 years cycle (<−2, −2~2, 2+ pounds), baseline high blood pressure and use of antihypertensive medications, depression, use of selective serotonin reuptake inhibitors (SSRI), use of other anti-depressive medications, hypercholesterolemia and use of lipid lowering medications, level of job demand and level of job control (in NHS and NHSII). In model 3, we further included average self-reported sleep duration, frequency of snoring, difficulty falling/staying asleep, and total number of years of having worked rotating night shifts (only assessed in NHS and NHS II, i.e. women only) as well as whether currently working rotating night shift in last 6 months (NHS only; assessed in 1996) in the estimation of probability of melatonin supplement use. Mode imputation was applied for variables with a missing rate < 5%; otherwise, participants with missing data were categorized into the “missing” group.
In secondary analyses, among women, we stratified the analyses by duration of rotating night shift work (no, 1–5 years, or >5 years). We tested the multiplicative interaction by comparing the −2 log likelihood of the multivariate adjusted models with and without the cross-product interaction term.21 To assess the additive interaction between not using melatonin and shift work, we considered duration of rotating night shift work as a continuous variable and assessed the relative excess risk due to interaction (RERI) as an index of additive interaction. 21
To test whether the associations with melatonin depended on the duration of supplement use, we defined the longer-term melatonin users as those who reported regular use of a melatonin supplement during eight years of follow-up (1998–2006 of NHS/HPFS and 2003–2011 of NHSII) for at least two cycles, and then examined its association with subsequent T2DM/CVD risk during follow-up from 2006 (NHS/HPFS) and 2011 (NHSII), respectively.
We applied the propensity score stratification method in all analysis. The propensity score (PS) was the estimated probability of melatonin supplement use conditional on observed covariates using a logistic regression. To test the robustness of our findings, we conducted several sensitivity analyses, including PS weighting method for main analysis, stratified analysis by age groups, and potential effect modification by melatonin receptor genetic variants (appendix).
We used SAS version 9.3 to analyze data. Statistical significance was set at a two-tailed P<0.05.
Results
The overall reported age-adjusted prevalence of melatonin use ranged between 1% to 2% from 1998 to 2007, with an increase beginning in 2008 in all 3 cohorts, which was 4.0% (95%CI: 3.7%−4.3%) in 2014 of HPFS and 5.3% (95%CI: 5.2%−5.5%) in NHSII in 2015 (Figure 1).
Figure 1. Age-standardized prevalence of regular use of melatonin supplements and time trends in the three cohorts.
(The %GLMCURV9 macro (https://ysph.yale.edu/cmips/research/software/glmcurv9/) using SAS PROC GENMOD and restricted cubic splines with 3 knots applying to flexibly model was used to test whether there was a nonlinear time trend in melatonin users standardized to the cohort-specific age distribution.)
For CVD analyses, we included 67,202 women in NHS, 65,241 women in NHSII and 26,629 men in HPFS who were free of CVD and cancers at baseline. Participants who reported regular use of melatonin supplements were more likely to report multivitamin use, aspirin use, coenzyme Q10 and fish oil supplement use, and post-menopausal hormone use (in women). These participants were also more likely to have a higher prevalence of hypercholesterolemia, hypertension, depression, diabetes and difficulty falling/staying asleep, and longer rotating night shift work duration in women, compared to those who did not report regular melatonin use (Table 1). Participants for T2DM analyses had similar differences in characteristics between melatonin users and non-users (appendix Table 1).
Table 1.
Baseline characteristics of the study population according to melatonin supplement use (NHS/HPFS 1998, NHSII 2003)1,2
| Regular use of melatonin supplement at baseline | ||||||
|---|---|---|---|---|---|---|
|
|
||||||
| No | Yes | No | Yes | No | Yes | |
| Baseline characteristics2 | NHS 1998 | NHS II 2003 | HPFS 1998 | |||
| N | 66,288 | 914 | 64,677 | 564 | 26,042 | 587 |
| Age (years), means (SD) | 63.6 (7.1) | 63.2 (7.0) | 48.2 (4.7) | 49.8 (4.3) | 62.9 (8.8) | 62.6 (8.4) |
| White, %(n) | 97.5 (64623) | 97.6 (892) | 94.7 (61264) | 96.5 (544) | 91.4 (23805) | 91.9 (539) |
| Hispanic, %(n) | 0.9 (588) | 1.7 (16) | 1.7 (1068) | 1.7 (10) | 0.7 (170) | 1.3 (8) |
| BMI (kg/m2), means (SD) | 26.6 (5.3) | 26.6 (5.4) | 27.0 (6.4) | 27.5 (6.9) | 26.1 (3.6) | 25.9 (3.7) |
| 2-year weight changes (lbs.), Median (IQR) | 0 (−3.0, 5.0) | 1.0(−3.0, 5.0) | 1.0 (−3.0, 6.0) | 1.0(−4.0, 7.0) | 0 (−3.0, 4.0) | 0 (−4.0, 5.0) |
| Physical activity (METs-h/w), Median (IQR) | 10.5(3.6,23. 4) | 10.2(3.7,23. 1) | 12.3(4.3,27. 4) | 12.3(4.3,30. 4) | 23.7(9.3,46. 2) | 28.4(12.7,50. 5) |
| Alcohol (g/day), Median (IQR) | 1.0 (0, 6.3) | 1.1 (0, 7.4) | 2.1 (0, 7.1) | 2.1 (0, 6.9) | 6.3 (0.9,15.1) | 6.9 (1.1,15.1) |
| AHEI, means (SD) | 48.5 (10.2) | 51.5 (11.4) | 51.5 (12.1) | 55.1 (13.3) | 50.0 (10.6) | 53.6 (11.1) |
| Smoking status, %(n) | ||||||
| Never smoking | 45.2 (29873) | 40.3 (368) | 66.1 (42740) | 57.8 (326) | 44.6 (11531) | 44.7 (260) |
| Former smokers | 44.3 (29312) | 48.3 (441) | 25.8 (16675) | 32.4 (183) | 50.6 (13069) | 52.3 (304) |
| Current smoking | 10.5 (6960) | 11.4 (104) | 7.9 (5124) | 9.8 (55) | 4.8 (1247) | 3.0 (18) |
| Multi-vitamin supplement use, %(n) | 60.7 (40241) | 77.1 (705) | 61.9 (40011) | 72.2 (407) | 56.8 (14787) | 79.2 (465) |
| Aspirin use, %(n) | 50.2 (33281) | 56.3 (514) | 11.6 (7511) | 18.9 (106) | 62.9 (16385) | 67.5 (396) |
| Coenzyme Q10 supplement, %(n) | 2.9 (1909) | 21.7 (198) | 3.4 (2217) | 19.3 (109) | 5.8 (1499) | 35.4 (208) |
| Fish oil supplement, %(n) | 2.0 (1312) | 13.2 (121) | 5.9 (3810) | 21.1 (119) | 3.5 (900) | 19.9 (117) |
| High job demand, %(n) | 40.1 (26587) | 42.6 (389) | 70.2 (45407) | 70.4 (397) | - | - |
| High job control, %(n) | 31.4 (20845) | 36.2 (331) | 54.2 (35071) | 52.9 (299) | - | - |
| Menopause status, %(n) | ||||||
| Premenopausal | 3.6 (2274) | 3.2 (28) | 68.0 (43920) | 63.5 (358) | - | - |
| Postmenopausal (No hormone use) | 25.0 (15866) | 15.1 (135) | 9.4 (6073) | 8.8 (50) | - | - |
| Postmenopausal (current hormone use) | 51.4 (32580) | 60.4 (540) | 12.9 (8314) | 15.1 (85) | - | - |
| Postmenopausal (former hormone use) | 20.0 (12667) | 21.3 (190) | 9.8 (6303) | 12.5 (71) | - | - |
| Family history, %(n) | ||||||
| Diabetes | 28.3 (18754) | 29.6 (270) | 41.0 (26524) | 43.5 (245) | 22.2 (5791) | 21.7 (127) |
| CVD | 25.4 (16832) | 26.2 (240) | 49.1 (31780) | 47.9 (270) | 36.6 (9530) | 35.6 (209) |
| Hypertension, %( n) | 46.7 (30977) | 52.6 (481) | 24.7 (15973) | 30.5 (172) | 36.9 (9611) | 43.0 (253) |
| Elevated cholesterol, %(n) | 56.1 (37183) | 60.8 (556) | 33.2 (21483) | 40.6 (229) | 45.2 (11782) | 50.1 (294) |
| Type 2 diabetes, %(n) | 6.7 (4409) | 8.4 (77) | 3.3 (2151) | 4.3 (24) | 5.5 (1442) | 6.9 (40) |
| Depression, %(n) | 6.5 (4298) | 11.7 (107) | 13.0 (8415) | 17.9 (101) | 12.7 (3317) | 18.8 (110) |
| Use of SSRI, %(n) | 5.7 (3770) | 10.7 (98) | 13.2 (8507) | 15.6 (88) | 2.4 (619) | 2.4 (14) |
| Use of other antidepressants, % (n) | 3.4 (2248) | 8.4 (77) | 8.3 (5374) | 12.8 (72) | 1.7 (438) | 3.0 (18) |
| Difficulty falling/staying asleep, %(n) | ||||||
| None of time | 35.0 (21166) | 17.7 (148) | 38.3 (22672) | 19.3 (99) | 72.6 (15524) | 60.4 (289) |
| Little of time | 31.6 (19106) | 27.8 (231) | 30.9 (18275) | 30.0 (153) | - | - |
| Some of time | 21.4 (12907) | 25.5 (212) | 17.4 (10319) | 25.9 (132) | 23.5 (5027) | 27.9 (134) |
| Good bit of time | 7.1 (4277) | 16.8 (140) | 7.7 (4537) | 13.0 (67) | - | - |
| Most of the time | 3.6 (2155) | 8.3 (69) | 4.3 (2535) | 8.0 (41) | 3.9 (845) | 11.7 (56) |
| All the time | 1.3 (813) | 4.0 (33) | 1.5 (867) | 3.7 (19) | - | - |
| Working rotating night shifts (years), %(n) | ||||||
| None | 41.3 (23150) | 37.4 (284) | 29.4 (19014) | 26.8 (151) | - | - |
| 1–5 years | 41.3 (23146) | 40.9 (311) | 51.0 (33002) | 52.8 (298) | - | - |
| 6–9 years | 6.5 (3648) | 8.4 (64) | 11.0 (7139) | 11.8 (66) | - | - |
| 10+ years | 10.9 (6088) | 13.3 (101) | 8.5 (5523) | 8.6 (48) | - | - |
| Rotating night Shift work in last 6 | 3.3 (2069) | 5.5 (48) | - | - | - | - |
| month, %(n) Sleep duration (hours) | ||||||
| <6 | 4.1 (2201) | 5.8 (43) | 5.4 (3188) | 5.7 (29) | 1.8 (350) | 3.3 (15) |
| 6–8 | 91.6 (49420) | 89.3 (664) | 89.5 (53144) | 85.6 (437) | 94.4 (17943) | 93.4 (418) |
| ≥9 | 4.4 (2350) | 4.9 (36) | 5.1 (3049) | 8.7 (44) | 3.8 (721) | 3.2 (14) |
| Frequency of snoring, %(n) | ||||||
| None | 26.9 (14511) | 28.6 (212) | 41.9 (21893) | 46.5 (212) | 25.9 (4881) | 24.7 (110) |
| Occasionally | 64.1 (34566) | 62.6 (462) | 40.1 (20948) | 34.6 (158) | 48.0 (9035) | 48.4 (215) |
| Frequently | 9.0 (4828) | 8.8 (65) | 17.9 (9356) | 18.9 (86) | 26.0 (4900) | 26.9 (119) |
NHS: Nurses’ Health Study; HPFS: Health Professionals Follow-up Study; MET: metabolic equivalent tasks; AHEI: Alternative Healthy Eating Index; CVD: cardiovascular diseases; SSRIs: selective serotonin reuptake inhibitors.
All participants are female in NHS and NHSII, and male in HPFS. Values are means (SD), median (interquartile range, IQR) or percentages (numbers) and were standardized to the age distribution of the study population except age.
Difficulty falling/staying asleep in HPFS was collected in 2004 with 3 frequencies; hypertension and elevated cholesterol in 3 cohorts, for all cycles, included either physician diagnosed or using medications. Depression was physician diagnosed, while the antidepressant medications were split into SSRI and other antidepressants.
We documented 16,917 incident CVD cases (9,218 in the NHS, 1,608 in the NHS II, and 6,091 in the HPFS) during 2,609,068 person-years of follow-up. In analyses pooling data from all three cohorts, the age- and sex- adjusted HR of CVD was 0.94 (95% confidence interval (CI): 0.84–1.07, P=0.35) comparing participants who reported regular use of melatonin supplements to non-users. The association was not materially changed with further adjustment for socioeconomic factors, lifestyle factors, baseline complications and other sleep related factors, with a pooled multivariable-adjusted HR of 0.94 (95%CI: 0.83–1.06, P=0.32) between regular use of melatonin supplement use and CVD, which was consistent in three cohorts (Table 2), among younger and older individuals, applying different propensity score analysis method, and among participants with different melatonin receptor genetic variants (appendix). Associations were also consistent across CVD subtypes, which was 1.04 (95%CI: 0.82–1.34, P=0.73) for CHD, 1.01 (95%CI: 0.79–1.29, P=0.91) for stroke and 0.86 (0.73–1.02, P=0.089) for coronary revascularization. The associations were not significant among those who reported melatonin supplement user in only one cycle (HR: 0.86; 95%CI: 0.72–1.03, P=0.093) nor among users in more than 1 cycles (HR: 1.11; 95%CI: 0.88–1.40, P=0.38). Compared to nurses who had no experience of rotating night shift work and did not report regular use of melatonin supplements (appendix Table 2), the hazard ratio of CVD among nurses who had 5+ years of shift work was 1.10 (95%CI: 1.04–1.17, P=0.0010) without regular use of melatonin supplements and 0.62 (95%CI: 0.39–0.97, P=0.036) with regular use of melatonin supplement (P-interaction=0.013 for both multiplicative and additive interactions) (Table 3).
Table 2.
Hazard ratios of cardiovascular diseases (CVD) and type 2 diabetes comparing participants with and without regular use of melatonin supplement
| No melatonin supplement use | Melatonin supplement use | HR (95% CI) comparing melatonin supplement users to non-users | |||||
|---|---|---|---|---|---|---|---|
|
|
|
|
|||||
| Cases | Person-years | Cases | Person-years | Model 1 | Model 2 | Model 3 | |
| CVD | |||||||
| NHS | 9,088 | 1,140,262 | 130 | 17,031 | 1.02 (0.86–1.21) P=0.84 | 1.04 (0.87–1.24) P=0.66 | 1.03 (0.87–1.23) P=0.74 |
| NHSII | 1,570 | 988,270 | 38 | 23,169 | 0.90 (0.65–1.25) P=0.53 | 0.82 (0.59–1.14) P=0.24 | 0.81 (0.59–1.13) P=0.21 |
| HPFS | 5,993 | 431,780 | 98 | 8,556 | 0.87 (0.71–1.06) P=0.17 | 0.89 (0.72–1.08) P=0.24 | 0.89 (0.73–1.09) P=0.25 |
| Pooled | 16,651 | 2,560,312 | 266 | 48,756 | 0.94 (0.84–1.07) P=0.35 | 0.95 (0.84–1.07) P=0.39 | 0.94 (0.83–1.06) P=0.32 |
| Type 2 Diabetes | |||||||
| NHS | 5,792 | 1,123,324 | 67 | 17,815 | 0.86 (0.68–1.09) P=0.22 | 0.88 (0.69–1.12) P=0.31 | 0.88 (0.69–1.12) P=0.29 |
| NHSII | 4,625 | 1,076,220 | 125 | 30,326 | 1.08 (0.90–1.29) P=0.41 | 1.01 (0.85–1.21) P=0.88 | 1.02 (0.85–1.03) P=0.84 |
| HPFS | 2,079 | 444,824 | 42 | 9,321 | 1.02 (0.75–1.39) P=0.88 | 1.07 (0.79–1.46) P=0.66 | 1.08 (0.79–1.47) P=0.64 |
| Pooled | 12,496 | 2,644,368 | 234 | 57,462 | 1.00 (0.87–1.13) P=0.95 | 0.99 (0.86–1.12) P=0.81 | 0.98 (0.86–1.12) P=0.80 |
Start of follow-up: NHS 1998, NHSII 2003, HPFS 1998.
End of follow-up: CVD: NHS/NHSII: 2019 HPFS 2020; Diabetes HPFS 2020/NHS 2021/NHSII 2023 Model 1: age- and sex- (for pooling) adjusted; Model 2: further adjusted race/ethnicity, use of aspirin, use of multi-vitamins, coenzyme Q10 supplement use, fish oil supplement use, family history of diabetes and CVD, smoking status, alcohol intake, AHEI and physical activity levels, body mass index at the beginning of each 2 years cycle and weight changes within each 2 years cycle, high blood pressure, using of anti-hypertensive medications, hypercholesterolemia, using of cholesterol lower medications, depression, use of use of selective serotonin reuptake inhibitors (SSRI), use of other anti-depressive medications, levels of job demand and levels of job control (in NHS and NHSII).
Model 3: further adjusted sleep duration, frequency of snoring, difficulty falling/staying asleep, and shift work duration (in NHS and NHSII).
To address the possibility of residual confounding of intent to treat, we applied the propensity score stratification method in both analyses of Model 2 and Model 3.
Table 3.
Joint association1 between regular use of melatonin supplement and rotating night shift work duration
| Regular use of melatonin supplement | No | Yes | No | Yes | No | Yes | |
|---|---|---|---|---|---|---|---|
| Duration of rotating night shift work | No | No | 1–5 years | 1–5 years | >5 years | >5 years | |
| CVD | Cases | 3,189 | 54 | 3,458 | 67 | 1,786 | 19 |
| Person years (PYs) | 665,563 | 11,928 | 862,192 | 17,021 | 365,696 | 7,941 | |
| IR NHS (/105 PYs) | 717 | 776 | 727 | 778 | 927 | 510 | |
| IR NHS II (/105 PYs) | 145 | 133 | 133 | 189 | 175 | 1112 | |
| Joint classification | HR1 (95%CI) | 1.0 (Ref.) | 1.04 (0.79–1.36) P=0.78 | 0.98 (0.93–1.03) P=0.44 | 1.11 (0.87–1.42) P=0.40 | 1.10 (1.04–1.17) P=0.0010 | 0.62 (0.39–0.97) P=0.036 |
| Stratified2 | HR1 (95%CI) | 1.0 (Ref.) | 1.05 (0.80–1.38) P=0.72 | 1.0 (Ref.) | 1.13 (0.89–1.44) P=0.32 | 1.0 (Ref.) | 0.55 (0.35–0.87) P=0.011 |
| HR for multiplicative interaction terms | 1.09 (0.76–1.56) P=0.65 | 0.54 (0.32–0.91) P=0.021 | |||||
| P for overall multiplicative interaction applying −2 log likelihood method with 2 degrees of freedom = 0.013 | |||||||
| P for additive interaction of per 5 years of shift work and not using melatonin = 0.013, with a RERI of 0.19 (95%CI: 0.04 – 0.33) | |||||||
| Diabetes | Cases | 3,081 | 48 | 4,201 | 87 | 2,250 | 43 |
| Person years (PYs) | 715,548 | 14,861 | 929,148 | 21,175 | 391,985 | 9,836 | |
| IR NHS (/105 PYs) | 478 | 322 | 502 | 366 | 618 | 330 | |
| IR NHS II (/105 PYs) | 369 | 324 | 414 | 430 | 545 | 478 | |
| Joint classification | HR1 (95%CI) | 1.0 (Ref.) | 0.90 (0.67–1.19) P=0.45 | 1.05 (1.00–1.10) P=0.037 | 1.10 (0.89–1.36) P=0.40 | 1.11 (1.05–1.17) P<0.001 | 0.91 (0.67–1.24) P=0.55 |
| Stratified2 | HR1 (95%CI) | 1.0 (Ref.) | 0.86 (0.65–1.15) P=0.32 | 1.0 (Ref.) | 1.08 (0.87–1.34) P=0.47 | 1.0 (Ref.) | 0.83 (0.61–1.13) P=0.23 |
| HR for multiplicative interaction terms | 1.17 (0.82–1.67) P=0.39 | 0.92 (0.61–1.39) P=0.69 | |||||
| P for overall multiplicative interaction3 applying −2 log likelihood method with 2 degrees of freedom = 0.39 | |||||||
| P for additive interaction of per 5 years of shift work and not using melatonin = 0.22, with a RERI of 0.08 (95%CI: −0.05 – 0.21) | |||||||
Based on pooled data from NHS and NHSII; IR: incident rate; HR: hazard ratio; CI: confident interval; RERI: relative excess risk due to interaction
End of follow-up: CVD: NHS/NHSII: 2019; Diabetes: NHS 2021/NHSII 2023
Adjusted for age, sex, race/ethnicity, use of aspirin, use of multi-vitamins, coenzyme Q10 supplement use, fish oil supplement use, family history of diabetes and CVD, smoking status, alcohol intake, AHEI, physical activity levels, body mass index at the beginning of each 2 years cycle and weight changes within each 2 years cycle, high blood pressure, using of anti-hypertensive medications, hypercholesterolemia, using of cholesterol lower medications, depression, use of selective serotonin reuptake inhibitors (SSRI), use of other anti-depressive medications, levels of job demand and levels of job control (in NHS and NHSII), sleep duration, frequency of snoring, and difficulty falling/staying asleep applying the propensity score stratification method.
HR between regular use of melatonin and CVD/diabetes among sub-groups of nurses stratified by shift work duration.
We documented 12,730 incident T2DM cases (5,859 in the NHS, 4,750 in the NHS II, and 2,122 in the HPFS) (Table 2). The multivariable-adjusted HR between regular use of melatonin supplement and T2DM was 0.98 (95%CI: 0.86–1.12, P=0.80), which was consistent across the three cohorts (Table 2); and consistent among short- and longer- term users of melatonin supplements: (HR: 0.89 (95%CI: 0.72–1.10, P=0.27) among those who reported melatonin use in only one cycle and 0.80 (95%CI: 0.57–1.12, P=0.19) among users in more than 1 cycles). Compared to nurses who had no experience of night rotating shift work and not regularly users of melatonin supplement, nurses with five or more years of rotating shift work experiences without regular melatonin supplement showed a 11% higher hazard of diabetes (HR=1.11, 95%CI: 1.05–1.17, P<0.001), which was 0.91 (95%CI: 0.67–1.24, P=0.55) among nurses with five or more years of shift work experience and also with regular users of melatonin supplement (P for multiplicative interaction=0.39 and P for additive interaction = 0.22) (Table 3).
Discussion
Based on two decades of follow-up of three large cohorts of middle-aged health care professionals, we did not observe significant associations between self-reported regular use of melatonin supplementation and risks of T2DM or CVD, regardless of other lifestyles and sleep conditions. However, there was some suggestive evidence that regular use of melatonin supplement might attenuate adverse effects of long-term rotating night shift work on T2DM and CVD risk among nurses.
Because endogenous melatonin, naturally produced under low-light conditions, is recognized as a key regulator of sleep–wake cycles, melatonin supplement use has become a popular choice for natural sleep aid. Previous studies suggested that melatonin might play a key role in reducing cardiovascular diseases through pathways that augment circadian rhythms.22 In addition to its fundamental role in promoting sleep, melatonin also has other regulatory roles in inflammation, immune system, oxidative stress and mitochondrial function.22 A previous meta-analysis based on clinical trial data suggested a significant association between melatonin supplementation and a reduction in triglycerides and total cholesterol level,23 and beneficial effects on fasting blood glucose, glycated hemoglobin, and insulin resistance.8 However, these meta-analyses were based on small studies with short follow-up among individuals with various diseases. Moreover, evidence from long-term randomized clinical trials or large cohort studies is lacking.8,24 Our results based on three large prospective studies do not support an overall association between regular use of melatonin supplements and risk of diabetes or CVD.
We observed a marginally significant inverse association of melatonin supplement with CVD risk among nurses who experienced long-term rotating night shift work. Endogenous melatonin secretion follows a circadian rhythm, typically peaking 3–5 hours after sleep onset when it is dark, with almost no production occurring during daylight and without any storage of melatonin in the body.25 Exposure to light at night, as with rotating night shift work, may disturb the body’s natural circadian rhythm and impair melatonin secretion. In the NHS II cohort, significantly lower urinary melatonin production during the night was observed among rotating-shift workers on night shifts compared to day-shift workers.26 Also, both night shift work27,28 and lower nocturnal melatonin secretion5–8 were prospectively and independently associated with increased risks for developing diabetes, hypertension, and myocardial infarction in these cohorts, suggesting that lower nocturnal melatonin secretion might be one mechanism connecting night shift work and chronic diseases. In our study, we observed significantly higher T2DM and CVD risk among nurses with 5 or more years of night shift work and without regular use of melatonin supplements, but no increased risk among those with regular use of melatonin supplements. However, the number of nurses with both long duration of night shift work and regular use of melatonin supplements was relatively small, resulting in inadequate statistical power for the subgroup analyses.
Our study has several strengths. First, it was based on three well-established cohorts with high rates of follow-up and well-validated assessment of cases, which minimized selection and ascertainment biases. The prospective design and long-term follow-up reduced the potential for reverse causation. Comprehensive information on demographics, lifestyle habits, and sleep - elated factors minimized the potential for residual confounding. Our study is, to our knowledge, the only prospective cohort study that evaluated the potential associations between regular use of melatonin supplements and long-term risk of T2DM and CVD.
Several limitations of this study should also be considered. First, this study was limited to melatonin use in predominantly white U.S. male and female health professionals. The relative homogeneity of our study participants improved the response rate and the quality of the self-reported health information and minimized confounding by socioeconomic status, allowing us to assess potential interactions between melatonin supplement use and shift work. However, this might lead to limited generalizability of our findings to other groups with different race/ethnicity and socioeconomic status. Compared to the national prevalence of 2.1% among adults with mean age of 47.5 years,3 the prevalence of melatonin supplement use was higher due to older age of our participants. Specifically, the prevalence of regular use of melatonin supplement is 4% among men with mean age of 77.6 years in 2014 and 5.1% of women with a mean age of 60.6 years in 2015. The relatively low prevalence of melatonin use limited our statistical power to detect a weak association, if it exists, which warrants future studies with larger sample sizes. The low number of melatonin supplement users also limited our power for in-depth analysis of variables that might modify the association between melatonin supplement use and health outcomes. Further, given the observational nature of the study, it is difficult to rule out the possibility of unmeasured confounding and residual confounding due to measurement errors in confounders, even though we carefully controlled for many potential confounders in the analyses, and the repeated measurements could potentially help to reduce the measurement errors in the confounders. We did not collect detailed information on melatonin supplements, such as age at initiation of use, frequency and duration of use, supplement dosage and brand, and whether the usage was related to night shift work. These limitations underscore the need for future studies in this area. Potential indication bias of melatonin use was also a concern as melatonin users tend to have sleep-related health conditions. To address this issue, we applied propensity score stratification method in all analyses to control for possibility of reverse causation bias.29 In addition, selection bias in estimating HRs in time-to-event analyses might occur30, although we did not observe significant violation of the proportional hazards assumption. Lastly, we refrained from drawing any causal inference from our findings due to the observational nature of the study.
In conclusion, based on the analyses of three well-established cohorts of U.S. men and women, we observed no significant association between regular use of melatonin supplements and risk of T2DM and CVD. Further research is warranted to assess if melatonin supplement use has the potential to mitigate the elevated risks of diabetes and CVD associated with rotating night shift work.
Supplementary Material
RESEARCH IN CONTEXT.
Evidence before this study
We searched PubMed for publications up to January 15, 2024, using the terms (melatonin supplements) AND ((cardiometabolic) OR (diabetes) OR (stroke) OR (heart) OR (cardiovascular)) in all fields. Previous studies indicated that use of melatonin supplements has significantly increased following its discovery 60 years ago, especially in the last two decades. The growing use of exogenous melatonin in the general population calls consideration of potential negative side effects. Although the current evidence for short-term cardiometabolic effects of melatonin supplements is relatively consistent, evidence on long-term risk of chronic diseases is lacking.
Added value of this study
Our study is, to our knowledge, the only prospective cohort study that evaluated the potential associations between regular use of melatonin supplements and long-term risk of type 2 diabetes (T2DM) and cardiovascular diseases (CVD). Based on two-decades follow-up of three large cohorts of middle-aged health care professionals, we documented 16,917 incident CVD cases and 12,730 incident T2DM cases. Self-reported melatonin supplement use was not significantly associated with the risk of T2DM or CVD. In secondary analyses, melatonin supplement use appeared to attenuate the positive association between long-term shift work (5+ years) and risk of CVD (P for interaction=0.013) among the nurses.
Implications of all the available evidence
This study has important clinical and public health implications. At the current melatonin supplement use level, we did not observe a significant association between regular use of the supplement and risk of T2DM and CVD in US adults. However, there was some suggestive evidence that regular use of melatonin supplement might attenuate adverse effects of long-term night shift work on CVD risk among nurses. Further research is warranted to assess if melatonin supplement use has the potential to mitigate the elevated risks of chronic diseases associated with rotating night shift work.
Acknowledgments
The authors would like to thank the participants and staff of the NHS and HPFS cohorts for their continuous involvement and valuable contributions. Special thanks to Ellen Hertzmark from Harvard School of Public Health for her help and guidance in the statistical analysis.
Sources of Funding
The research reported in this manuscript was supported by the National Institutes of Health grants UM1 CA186107, U01 CA176726, U01 CA167552, U01 HL145386, R01 HL034594, R01 HL088521, and R01 HL35464. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Disclosures: Redline S received consulting fees from Eli Lilly, Apnimed, and Jazz Pharma and was on the board for Alliance for Sleep Apnea Partners (no support), National Sleep Foundation (no support), and NHLBI Council. All others declared none to disclose.
Data Sharing Statement: Because of participant confidentiality and privacy concerns, data cannot be shared publicly and requests to access NHS/NHSII/HPFS data must be submitted in writing. According to standard controlled access procedures, applications to use NHS/NHSII/HPFS resources will be reviewed by our External Collaborations Committee to verify that the proposed use maintains the protection of the privacy of participants and the confidentiality of the data. Investigators wishing to use NHS/NHSII/HPFS data are asked to submit a brief description of the proposed project (go to https://www.nurseshealthstudy.org/researchers (contact email: nhsaccess@channing.harvard.edu) and https://sites.sph.harvard.edu/hpfs/for-collaborators/ for details.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Besag FMC, Vasey MJ, Lao KSJ, Wong ICK. Adverse events associated with melatonin for the treatment of primary or secondary sleep disorders: a systematic review. CNS Drugs. 2019; 33(12): 1167–1186. [DOI] [PubMed] [Google Scholar]
- 2.Kuehn BM. Climbing melatonin use for insomnia raises safety concerns. JAMA. 2022; 328(7): 605–607. [DOI] [PubMed] [Google Scholar]
- 3.Li J, Somers VK, Xu H, Lopez-Jimenez F, Covassin N. Trends in use of melatonin supplements among US adults, 1999–2018. JAMA 2022; 327(5): 483–485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Black LI, Clarke TC, Barnes PM, Stussman BJ, Nahin RL. Use of complementary health approaches among children aged 4–17 years in the United States: National Health Interview Survey, 2007–2012. Natl Health Stat Report. 2015; 78: 1–19. [PMC free article] [PubMed] [Google Scholar]
- 5.Forman JP, Curhan GC, Schernhammer ES. Urinary melatonin and risk of incident hypertension among young women. J Hypertens 2010; 28(3): 446–451. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.McMullan CJ, Curhan GC, Schernhammer ES, Forman JP. Association of nocturnal melatonin secretion with insulin resistance in nondiabetic young women. Am J Epidemiol. 2013; 178(2): 231–238 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.McMullan CJ, Schernhammer ES, Rimm EB, Hu FB, Forman JP. Melatonin secretion and the incidence of type 2 diabetes. JAMA 2013; 309(13): 1388–1396. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.McMullan CJ, Rimm EB, Schernhammer ES, Forman JP. A nested case-control study of the association between melatonin secretion and incident myocardial infarction. Heart. 2017; 103(9): 694–701. [DOI] [PubMed] [Google Scholar]
- 9.Rubio-Sastre P, Scheer FAJL, Gómez-Abellán P, Madrid JA, Garaulet M. Acute melatonin administration in humans impairs glucose tolerance in both the morning and evening. Sleep 2014; 37(10): 1715–1719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Cagnacci A, Arangino S, Renzi A, et al. Influence of melatonin administration on glucose tolerance and insulin sensitivity of postmenopausal women. Clin Endocrinol 2001; 54(3): 339–346. [DOI] [PubMed] [Google Scholar]
- 11.Kampmann U, Lauritzen ES, Grarup N, et al. Acute metabolic effects of melatonin - a randomized crossover study in healthy young men. J Pineal Res. 2021; 70(2): e12706. [DOI] [PubMed] [Google Scholar]
- 12.Garfinkel D, Zorin M, Wainstein J, Matas Z, Laudon M, Zisapel N. Efficacy and safety of prolonged-release melatonin in insomnia patients with diabetes: a randomized, double-blind, crossover study. Diabetes Metab Syndr Obes 2011; 4: 307–313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Rechciński T, Trzos E, Wierzbowska-Drabik K, Krzemińska-Pakuła M, Kurpesa M. Melatonin for nondippers with coronary artery disease: assessment of blood pressure profile and heart rate variability. Hypertens Res 2010; 33: 56–61. [DOI] [PubMed] [Google Scholar]
- 14.Delpino FM, Figueiredo LM, Nunes BP. Effects of melatonin supplementation on diabetes: A systematic review and meta-analysis of randomized clinical trials. Clinical Nutrition 2021; 40: 4595e4605. [DOI] [PubMed] [Google Scholar]
- 15.Jiki Z, Lecour S, Nduhirabandi F. Cardiovascular benefits of dietary melatonin: A myth or a reality? Front Physiol 2018; 9: 528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chiuve SE, Fung TT, Rimm EB, et al. Alternative dietary indices both strongly predict risk of chronic disease. J Nutr. 2012;142(6):1009–1018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Bao Y, Bertoia ML, Lenart EB, et al. Origin, methods, and evolution of the three Nurses’ Health Studies. Am J Public Health. 2016;106(9):1573–1581. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Karasek R, Theorell T. Healthy Work: Stress, Productivity and the Reconstruction of Working Life. Basic Books; 1990. [Google Scholar]
- 19.National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes. 1979;28(12):1039–1057 [DOI] [PubMed] [Google Scholar]
- 20.Shan Z, Li Y, Baden MY et al. Association between healthy eating patterns and risk of cardiovascular disease. JAMA Intern Med. 2020;180(8):1090–1100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.VanderWeele TJ, Knol MJ. A tutorial on interaction. Epidemiol Meth 2014; 3: 33–72. [Google Scholar]
- 22.Zhong J, Liu Y. Melatonin and age-related cardiovascular diseases. Aging Medicine. 2018; 1(2): 197–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Mohammadi-Sartang M, Ghorbani M, Mazloom Z. Effects of melatonin supplementation on blood lipid concentrations: A systematic review and meta-analysis of randomized controlled trials. Clin Nutr. 2018;37(6 Pt A):1943–1954. [DOI] [PubMed] [Google Scholar]
- 24.Martín Giménez VM, de Las Heras N, Lahera V, Tresguerres JAF, Reiter RJ, Manucha W. Melatonin as an anti-aging therapy for age-related cardiovascular and neurodegenerative Diseases. Front Aging Neurosci. 2022;14:888292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Peschke E, Mühlbauer E. New evidence for a role of melatonin in glucose regulation. Best Pract Res Clin Endocrinol Metab 2010;24(5):829–841. [DOI] [PubMed] [Google Scholar]
- 26.Razavi P, Devore EE, Bajaj A, et al. Shift work, chronotype, and melatonin rhythm in nurses. Cancer Epidemiol Biomarkers Prev. 2019; 28(7): 1177–1186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Shan Z, Li Y, Zong G, et al. Rotating night shift work and adherence to unhealthy lifestyle in predicting risk of type 2 diabetes: results from two large US cohorts of female nurses. BMJ. 2018;363:k4641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Vetter C, Devore EE, Wegrzyn LR, et al. Association between rotating night shift work and risk of coronary heart disease among women. JAMA. 2016;315(16):1726–1734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Austin PC. An Introduction to Propensity Score Methods for Reducing the Effects of Confounding in Observational Studies. Multivariate Behav Res. 2011; 46(3): 399–424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hernán MA. The hazards of hazard ratios. Epidemiology 2010; 21:13–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.