Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Mar 1.
Published in final edited form as: Am J Prev Med. 2015 Jan 6;48(3):241–252. doi: 10.1016/j.amepre.2014.10.018

Total and Cause-Specific Mortality of U.S. Nurses Working Rotating Night Shifts

Fangyi Gu 1, Jiali Han 1, Francine Laden 1, An Pan 1, Neil E Caporaso 1, Meir J Stampfer 1, Ichiro Kawachi 1, Kathryn M Rexrode 1, Walter C Willett 1, Susan E Hankinson 1, Frank Speizer 1, Eva S Schernhammer 1
PMCID: PMC4339532  NIHMSID: NIHMS640388  PMID: 25576495

Abstract

Background

Rotating night shift work imposes circadian strain and is linked to the risk of several chronic diseases.

Purpose

To examine associations between rotating night shift work and all-cause, cardiovascular disease (CVD), and cancer mortality in a prospective cohort study of 74,862 registered U.S. nurses from the Nurses’ Health Study.

Methods

Lifetime rotating night shift work (defined as ≥3 nights/month) information was collected in 1988. During 22 years (1988–2010) of follow-up, 14,181 deaths were documented, including 3,062 CVD and 5,413 cancer deaths. Cox proportional hazards models (2013) estimated multivariable-adjusted hazard ratios (HRs) and 95% CIs.

Results

All-cause and CVD mortality were significantly increased among women with ≥5 years of rotating night shift work, compared to women who never worked night shifts. Specifically, for women with 6–14 and ≥15 years of rotating night shift work, the HRs were 1.11 (95% CI=1.06, 1.17) and 1.11 (95% CI=1.05, 1.18) for all-cause mortality and 1.19 (95% CI=1.07, 1.33) and 1.23 (95% CI=1.09, 1.38) for CVD mortality. There was no association between rotating night shift work and all-cancer mortality (HR≥15years=1.08, 95% CI=0.89, 1.19) or any other cancer, with the exception of lung cancer (HR≥15years=1.25, 95% CI=1.04, 1.51).

Conclusions

Women working rotating night shifts for ≥5 five years have a modest increase in all-cause and CVD mortality; those working ≥15 years of rotating night shift work have a modest increase in lung cancer mortality. These results add to prior evidence of a potentially detrimental effect of rotating night shift work on health and longevity.

Introduction

Shift work including a nighttime rotation is common, with up to 20% of the western workforce encountering alternate work schedules some time during their career.1 Observational studies consistently suggest positive associations of night shift work and cardiovascular disease (CVD)2 and cancer risk.3-18 Thus, the WHO in 2007 classified night work as a probable carcinogen.19 In addition to the link with cancer and CVD risk, other health outcomes including diabetes, hypertension, chronic fatigue, various sleep problems,20-23 and higher body weight22 are associated with shift work.24-26 However, evidence is more limited and inconsistent27 on associations between night shift work and all-cause and cause-specific mortality in observational studies.18,28-32

The circadian system and its prime marker, melatonin, are considered to have antitumor effects through multiple pathways, including antioxidant activity, anti-inflammatory effects, and immune enhancement.33-37 They also exhibit beneficial actions on cardiovascular health by enhancing endothelial function, maintaining metabolic homeostasis,38 and reducing inflammation.39,40 Direct nocturnal light exposure suppresses melatonin production41 and resets the timing of the circadian clock.33,42,43 In addition, sleep disruption may also accentuate the negative effects of night work on health.44,45 Taken together, substantial biological evidence supports the role of night shift work in the development of poor health conditions including cancer, CVD, and ultimately, mortality.

The current analysis is based on 22 years of follow-up of 74,862 women participating in the Nurses’ Health Study (NHS). This large U.S. cohort of nurses provides a unique opportunity to study associations between duration of rotating night shift work, all-cause, and cause-specific mortality. The baseline age range (30–55 years), high death rate, and nearly complete follow-up allows for a powerful analysis of night shift work and mortality, with repeat measures available on a number of important confounding factors. The 1988 assessment of lifetime night work likely captures most of the night shift work history of women retained in this cohort, as, based on data from the NHS2 cohort,46 <5% of nurses continue to work night shifts in middle age and <2% commence working night shifts at this age.

Methods

Study Population

The NHS cohort was established in 1976 when 121,701 U.S. female registered nurses aged 30–55 years were enrolled. All women completed an initial questionnaire at baseline about their lifestyle factors, health behaviors, and medical history, and have since been followed using biennial questionnaires to update exposure status and disease diagnoses.47 Night shift work information was collected in 1988. Of the 103,613 nurses who responded to the 1988 questionnaire, 85,197 (82.2%) answered the question about shift work. These women did not differ substantially from non-respondents in terms of their risk profile.48 After excluding women with pre-existing CVD (n=2,444) or any cancer other than non-melanoma skin cancer in or before 1988 (n=7,891), 74,862 women were included in the analytic sample. The study protocol was approved by the IRBs of Brigham and Women’s Hospital and Harvard School of Public Health (Boston, MA).

Measures

Information on lifetime rotating night shift work was collected once in 1988. On the 1988 questionnaire, participants were asked What is the total number of years during which you worked rotating night shifts at least 3 nights/month in addition to days or evenings in that month?, with eight prespecified response categories: never, 1–2, 3–5, 6–9, 10–14, 15–19, 20–29, and ≥30 years. This definition of rotating night shifts (i.e., ≥3 nights/month) did not specifically distinguish between rotating or permanent night shifts.

Deaths in the cohort were identified by reports from next of kin and via postal authorities. In addition, names of non-responders were searched in the National Death Index, which is highly sensitive and specific in this cohort.49 The underlying cause of death was determined by physician review primarily based on medical records and death certificates. ICD-8 was used to distinguish deaths due to cancer (ICD codes 140-207) and CVD (ICD codes 390–459 and 795). Cause-specific death from CVD and cancer sites with >100 deaths included in analysis were: ischemic heart disease (IHD; 410–414), cerebrovascular disease (430–438), and any other CVD (390–459 and 795 excluding 410–414 and 430–438); cancers of the lung (162), breast (174), ovaries (183), pancreas (157), colon or rectum (153), brain (191), kidney (189), uterus (182); and non-Hodgkin’s lymphoma (200, 202, 204), myeloma (203), leukemia (205–207), and other cancers (199). Follow-up rates for this cohort remain at approximately 90%, and the overall death follow-up is >98% complete.49

Statistical Analysis

Age and age-adjusted demographic and lifestyle factors were summarized at the midpoint of follow-up (1998) according to night shift work categories (never, 1–5, 6–14, ≥15 years). Cox proportional hazards regression models were used (2013) to estimate hazard ratios (HRs) and 95% CIs were calculated for all-cause mortality, CVD and cancer mortality, and mortality from specific CVD and cancer sites with >100 deaths. Person-years of follow-up were accrued from baseline (June 1988) to the date of death, loss to follow-up, or June 2010, whichever came first. Tests for linear trend across categories of night shift work were performed by assigning the median value of each shift work category. Each endpoint was adjusted for age and regularly updated information throughout follow-up on: alcohol consumption, physical exercise, multivitamin use, menopausal status and postmenopausal hormone use, physical examination in the past 2 years, healthy eating score37 (without components of multivitamin and alcohol, which were adjusted individually), smoking status, pack-years, and BMI. Information on these covariates was updated at 2–4-year intervals throughout follow-up. Additional adjustment for husbands’ education (less than high school, some high school, high school graduate, college, graduate school) was performed as the best indicator of SES; further adjustment for nurses’ education left the results virtually unchanged and was therefore not retained in the final models. Sensitivity analyses restricted to women with sleep duration of 6–8 hours were performed. Potential intermediate covariates (e.g., type 2 diabetes (T2D), hypertension, and hypercholesterolemia) were added to multivariable models. The proportional hazards assumption was tested by including interaction terms between duration of night shift work categories and follow-up time.

As smoking and obesity are well-established risk factors for CVD and multiple cancers, stratified analyses were performed by smoking status (never, former, current) and BMI (18.5–24.9, 25.0–29.9, ≥30.0 kg/m2). Interaction p-values were calculated by adding interaction terms into the models and performing likelihood ratio tests. To reduce residual confounding, adjustment for pack-years among former and current smokers, and continuous BMI within each BMI stratum, were performed.

Results

Over 22 years and 1.5 million person-years of follow-up, 14,181 deaths were documented, of which 3,062 were attributed to CVD and 5,413 to cancer. Women who had never worked rotating night shifts accounted for 41% of the person-years of follow up, and those who worked 1–5, 6–4, and ≥15 years on shifts accounted for 41%, 11%, and 7% of person-years, respectively. Table 1 shows that women with longer durations of rotating night shift work tended to be older (mean age, 66.3 years for ≥15 years of shift work vs 63.6–64.6 years for others) and had a higher BMI and were more physically active after standardizing for age. They were also more likely to be current smokers, less likely to use postmenopausal hormones or multivitamins, and of those married, the husbands tended to be less educated. There were no appreciable differences in dietary factors across durations of shift work; however, women who worked night shifts for longer durations tended to drink less alcohol and ate less daily cereal fiber compared to women without night shift work. They also had gained more weight since age 18 years, and were more likely to have a history of diabetes, hypertension, and hypercholesterolemia.

Table 1.

Age-standardized characteristics of participants in the NHS cohort in 1998, by baseline night shifts (n=71,857).

Night shift work duration at baseline in 1988
Never (n=29,328) 1–5 years (n=29,411) 6–14 years (n=8,087) ≥15 years (n=5,031)
Characteristics % Mean (SD) % Mean (SD) % Mean (SD) % Mean (SD)
Participants (%) 41 41 11 7
Current smoker 9 10 13 14
Past smoker 45 46 45 43
Physical exam 92 92 92 92
Postmenopausal women 94 94 94 93
Postmenopausal hormone therapy 50 50 47 43
Current multivitamin use 63 63 62 60
Family history of breast cancer 14 14 14 14
Benign breast disease 20 20 20 18
Husband’s highest education, college
and graduate school		 55 57 50 43
Diabetes mellitus 7 7 10 12
Hypertension 44 44 47 50
Hypercholesterolemia 57 57 58 58
Age in 1998, y 63.6 (7.1) 63.8 (7.1) 64.6 (7.1) 66.3 (6.8)
Body mass index, kg/m2 26.4 (5.2) 26.5 (5.3) 27.4 (5.6) 28.0 (5.8)
Physical exercise, met-h/wk 16.6 (19.9) 17.9 (21.6) 17.8 (21.8) 18.1 (27.3)
Walking, met-h/wk 6.2 (8.4) 6.6 (8.6) 6.6 (9.0) 6.4 (9.2)
Pack-years 24.1 (21.4) 23.7 (21.0) 26.4 (21.8) 27.0 (21.7)
Alcohol, g/d 5.1 (9.2) 5.2 (9.2) 4.5 (8.7) 3.9 (7.9)
Red meat, serving/d 0.8 (0.4) 0.8 (0.4) 0.8 (0.4) 0.8 (0.4)
Vegetables, serving/d, 3.1 (1.3) 3.3 (1.4) 3.3 (1.4) 3.3 (1.5)
Fruit, serving/d 2.3 (1.1) 2.4 (1.2) 2.4 (1.2) 2.3 (1.2)
Polyunsaturated fat, g/d 10.8 (5.1) 10.9 (4.9) 10.8 (5.0) 10.8 (5.0)
Saturated fat, g/d 17.9 (8.0) 18.1 (8.1) 18.4 (8.3) 18.7 (8.4)
Trans fat, g/d 3.0 (1.5) 3.1 (1.5) 3.1 (1.6) 3.1 (1.6)
Cereal fiber, g/d 6.8 (3.8) 6.8 (3.8) 6.6 (3.7) 6.5 (3.6)
Height, Inch 64.5 (2.4) 64.6 (2.4) 64.5 (2.5) 64.5 (2.5)
Age at first birth, y 24.5 (3.7) 24.9 (3.7) 25.1 (4.6) 24.6 (3.9)
Weight change from age 18, 1b 13.9 (12.9) 14.1 (13.0) 15.3 (14.0) 16.5 (14.6)
Open in a new tab

d, day; g, gram; h, hour; kg/m, kilogram per meter; lb, pound; met, metabolic equivalent value; wk, week; y, year.

Age- and multivariable-adjusted HRs for the associations between night shift work and all-cause and cause-specific mortality are presented in Table 2. Women working rotating night shifts for >5 years had modest increases in both all-cause and CVD-related mortality, compared to women who never worked rotating night shifts. For all-cause mortality, the multivariable-adjusted HRs for women with 1–5, 6–14, or ≥15 years of rotating night shift work were 1.01 (95% CI=0.97, 1.05), 1.11 (95% CI=1.06, 1.17), and 1.11 (95% CI=1.05, 1.18; ptrend<0.001). For CVD mortality, the multivariable-adjusted HRs were 1.02 (95% CI=0.94, 1.11), 1.19 (95% CI=1.07, 1.33), and 1.23 (95% CI=1.09, 1.38; ptrend<0.001). Women working night shifts for 6–14 or ≥15 years also had higher risk of all-cancer mortality. Compared to women who never worked rotating night shifts, the age-adjusted HRs were 1.10 (95% CI=1.01, 1.20) and 1.20 (95% CI=1.09, 1.32), but these associations were attenuated after multivariable adjustment, with HRs of 1.04 (95% CI=0.95, 1.13) and 1.08 (95% CI=0.98, 1.19), respectively.

Table 2.

Association between night shift work and all cause, CVD, and cancer mortality in NHS, 1988-2010 (n=74,862).

Night shift work duration
Mortality Never 1–5 years 6–14 years ≥15 years p for trend
All causes
 No. of deaths 5,417 5,424 1,910 1,430
 Age-adjusted HR (95% CI) 1.00 (ref) 0.97 (0.94, 1.01) 1.19 (1.13, 1.25) 1.24 (1.17, 1.32) <0.001
 Multivariable HR (95% CI)a 1.00 (ref) 1.01 (0.97, 1.05) 1.11 (1.06, 1.17) 1.11 (1.05, 1.18) <0.001
All cardiovascular disease
 No. of deaths 1,128 1,128 442 364
 Age-adjusted HR (95% CI) 1.00 (ref) 0.97 (0.90, 1.06) 1.30 (1.16, 1.45) 1.45 (1.29, 1.63) <0.001
 Multivariable HR (95% CI)a 1.00 (ref) 1.02 (0.94, 1.11) 1.19 (1.07, 1.33) 1.23 (1.09, 1.38) <0.001
All cancer
 No. of deaths 2,087 2,148 672 506
 Age-adjusted HR (95% CI) 1.00 (ref) 1.00 (0.94, 1.06) 1.10 (1.01, 1.20) 1.20 (1.09, 1.32) <0.001
 Multivariable HR (95% CI)a 1.00 (ref) 1.03 (0.97, 1.09) 1.04 (0.95, 1.13) 1.08 (0.98, 1.19) 0.11
Open in a new tab

Boldface indicates statistical significance (p<0.05).

HR, hazard ratio; NHS, Nurses’ Health Study; CVD, Cardiovascular diseases.

a

Multivariable model adjusted for age (continuous), alcohol consumption (none, 0.1-4.9, 5.0-14.9, ≥15.0 g/d), physical exercise (metabolic equivalent values; quintiles), multivitamin use (yes, no), menopausal status (premenopausal, postmenopausal) and postmenopausal hormone use (never, past and current), physical exam in the past 2 years (no , yes for symptoms and yes for screenings), healthy eating score (quintiles), smoking status (never, former, current), pack-years (<10, 10-19, 20-39, ≥40 for former smokers; <25, 25-44, 45-64, ≥65 for current smokers), and BMI (kg/m2; <21, 21-22.9, 23-24.9, 25-27.4, 27.5-29.9, 30-34.9, ≥35) and husband’s education (less than high school, some high school, high school graduate, college, graduate school, missing category).

Results for specific cancer sites with ≥200 deaths are offered in Table 3. For lung cancer mortality, the age-adjusted HR for ≥15 years of rotating night shift work was 1.44 (95% CI=1.20, 1.73), and 1.25 after multivariate adjustment (95% CI=1.04, 1.51), with smoking as the driving confounder. For colorectal cancer, the age-adjusted HR after ≥15 years of rotating night shift work was 1.42 (95% CI=1.04, 1.94), but was attenuated to 1.33 (95% CI=0.97, 1.83) after multivariate adjustment. Adjusting for current disease status (T2D, hypertension, and hypercholesterolemia) left results largely unchanged (data not shown). Although no increase in breast cancer mortality was observed among women with ≥15 years of rotating night shift work (multivariable HR=0.99, 95% CI=0.74, 1.33), among women with ≥30 years of rotating night shift work, the multivariable HR for breast cancer mortality was 1.47 (95% CI=0.94, 2.32). Cancers with <200 deaths are presented in Appendix Table 1. There was a non-significant increase in mortality from kidney cancer and myeloma among women with 6–14 and ≥15 years of night work: the multivariable HRs were 1.72 (95% CI=1.03, 2.86) and 1.39 (95% CI=0.75, 2.57) for kidney cancer (ptrend=0.048) and 1.56 (95% CI=0.93, 2.64) and 1.61 (95% CI=0.90, 2.88) for myeloma (ptrend=0.08). There was no significant association between rotating night shift work and mortality from other cancers.

Table 3.

Association between night shift work and cancer-specific (≥ 200 deaths) mortality in NHS, 1988-2010 (n=74,862).

Night shift work duration
Mortality Never 1–5 years 6–14 years ≥15 years p for trend
Lung cancer
No. of deaths 501 523 168 150
Age-adjusted HR (95% CI) 1.00 (ref) 1.02 (0.91, 1.16) 1.14 (0.96, 1.36) 1.44 (1.20, 1.73) <0.001
Multivariable HR (95% CI)a 1.00 (ref) 1.05 (0.92, 1.19) 0.99 (0.83, 1.18) 1.25 (1.04, 1.51) 0.05
Breast cancer
No. of deaths 269 293 79 55
Age-adjusted HR (95% CI) 1.00 (ref) 1.05 (0.89, 1.25) 1.04 (0.81, 1.34) 1.08 (0.80, 1.44) 0.64
Multivariable HR (95% CI)a 1.00 (ref) 1.07 (0.90, 1.26) 0.99 (0.76, 1.27) 0.99 (0.74, 1.33) 0.83
Ovarian cancer
No. of deaths 180 168 47 30
Age-adjusted HR (95% CI) 1.00 (ref) 0.92 (0.74, 1.13) 0.90 (0.65, 1.24) 0.86 (0.58, 1.27) 0.41
Multivariable HR (95% CI)a 1.00 (ref) 0.95 (0.77, 1.18) 0.89 (0.64, 1.23) 0.82 (0.55, 1.22) 0.27
Pancreatic cancer
No. of deaths 149 173 52 33
Age-adjusted HR (95% CI) 1.00 (ref) 1.12 (0.90, 1.39 1.20 (0.88, 1.65) 1.10 (0.75, 1.60) 0.47
Multivariable HR (95% CI)a 1.00 (ref) 1.12 (0.90, 1.40) 1.14 (0.83, 1.58) 1.03 (0.70, 1.51) 0.77
Colorectal cancer
No. of deaths 180 176 56 52
Age-adjusted HR (95% CI) 1.00 (ref) 0.95 (0.77, 1.17) 1.07 (0.80, 1.45) 1.42 (1.04, 1.94) 0.02
Multivariable HR (95% CI)a 1.00 (ref) 0.98 (0.79, 1.21) 1.05 (0.77, 1.42) 1.33 (0.97, 1.83) 0.07
Non Hodgkin’s Lymphoma
No. of deaths 103 89 34 26
Age-adjusted HR (95% CI) 1.00 (ref) 0.85 (0.66, 1.09) 0.94 (0.65, 1.37) 1.11 (0.74, 1.66) 0.57
Multivariable HR (95% CI)a 1.00 (ref) 0.87 (0.67, 1.12) 0.96 (0.66, 1.40) 1.06 (0.71, 1.58) 0.77
Other cancer (ICD=199)
No. of deaths 146 145 35 32
Age-adjusted HR (95% CI) 1.00 (ref) 0.95 (0.76, 1.20) 0.80 (0.55, 1.16) 1.05 (0.72, 1.55) 0.87
Multivariable HR (95% CI)a 1.00 (ref) 0.98 (0.77, 1.23) 0.75 (0.52, 1.09) 0.94 (0.63, 1.38) 0.39
Open in a new tab

Boldface indicates statistical significance (p<0.05).

HR, hazard ratio; NHS, Nurses’ Health Study.

a

Multivariable model adjusted for age (continuous), alcohol consumption (none, 0.1-4.9, 5.0-14.9, ≥15.0 g/d), physical exercise (metabolic equivalent values; quintiles), multivitamin use (yes, no), menopausal status (premenopausal, postmenopausal) and postmenopausal hormone use (never, past and current), physical exam in the past 2 years (no, yes for symptoms and yes for screenings), healthy eating score (quintiles), smoking status (never, former, current), pack-years (<10, 10-19, 20-39, ≥40 for former smokers; <25, 25-44, 45-64, ≥65 for current smokers), BMI (kg/m2; <21, 21-22.9, 23-24.9, 25-27.4, 27.5-29.9, 30-34.9, ≥35) and husband’s education (less than high school, some high school, high school graduate, college, graduate school, missing category).

For mortality from specific types of CVD with >100 deaths (Table 4), increased IHD mortality was observed among women who worked rotating night shifts for 6–14 and ≥15 years (HR=1.22, 95% CI=1.02, 1.46; 1.31, 1.09, 1.59, respectively; ptrend<0.001). Additional adjustment for current disease status (T2D, hypertension, and hypercholesterolemia) only minimally affected these associations (Table 4). There were no significant associations among women working rotating night shifts for 6–14 and ≥15 years for cerebrovascular disease or any other CVD.

Table 4.

Association between night shift work and CVD-specific (>100 deaths) mortality in NHS, 1988-2010 (n=74,862).

Night shift work duration
Mortality Never 1–5 years 6–14 years ≥15 years p for trend
Ischaemic heart disease
No. of deaths 407 386 170 153
Age-adjusted HR (95% CI) 1.00 (ref) 0.92 (0.80, 1.05) 1.38 (1.15, 1.65) 1.69 (1.40, 2.04) <0.001
Multivariable HR (95% CI)a 1.00 (ref) 0.97 (0.85, 1.12) 1.23 (1.03, 1.47) 1.34 (1.11, 1.61) <0.001
Multivariable HR (95% CI)b 1.00 (ref) 0.96 (0.83, 1.11) 1.17 (0.98, 1.41) 1.23 (1.02, 1.49) 0.006
Cerebrovascular disease
No. of deaths 319 320 119 90
Age-adjusted HR (95% CI) 1.00 (ref) 0.97 (0.83, 1.14) 1.24 (1.00, 1.53) 1.23 (0.97, 1.55) 0.02
Multivariable HR (95% CI)a 1.00 (ref) 1.02 (0.87, 1.20) 1.20 (0.97, 1.48) 1.12 (0.88, 1.42) 0.17
Multivariable HR (95% CI)b 1.00 (ref) 1.01 (0.86, 1.18) 1.16 (0.94, 1.44) 1.06 (0.83, 1.34) 0.36
Other cardiovascular disease
No. of deaths 402 422 153 121
Age-adjusted HR (95% CI) 1.00 (ref) 1.02 (0.89, 1.16) 1.28 (1.06, 1.54) 1.36 (1.11, 1.67) <0.001
Multivariable HR (95% CI)a 1.00 (ref) 1.05 (0.91, 1.20) 1.18 (0.97, 1.42) 1.17 (0.95, 1.44) 0.06
Multivariable HR (95% CI)b 1.00 (ref) 1.04 (0.90, 1.19) 1.14 (0.94, 1.38) 1.13 (0.92, 1.39) 0.14
Open in a new tab

Boldface indicates statistical significance (p<0.05).

HR, hazard ratio; NHS, Nurses’ Health Study; CVD, Cardiovascular diseases

a

Multivariable model adjusted for age (continuous), alcohol consumption (none, 0.1-4.9, 5.0-14.9, ≥15.0 g/d), physical exercise (metabolic equivalent values; quintiles), multivitamin use (yes, no), menopausal status (premenopausal, postmenopausal) and postmenopausal hormone use (never, past and current), physical exam in the past 2 years (no , yes for symptoms and yes for screenings), healthy eating score (quintiles), smoking status (never, former, current), pack-years (<10, 10-19, 20-39, ≥40 for former smokers; <25, 25-44, 45-64, ≥65 for current smokers), BMI (kg/m2; <21, 21-22.9, 23-24.9, 25-27.4, 27.5-29.9, 30-34.9, ≥35) and husband’s education (less than high school, some high school, high school graduate, college, graduate school, missingness).

b

This model was further adjusted for hypertension (yes, no), hypercholesterolemia (yes, no), type 2 diabetes (yes, no),

When stratifying by smoking status (Table 5), the association of rotating night shift work appeared to be stronger among current smokers for all-cause mortality (HR≥15 years, 1.21 vs 1.11), but remained statistically significant also among never smokers (ptrend=0.05). Similarly, for CVD and IHD mortality, HRs were comparable among never smokers and the entire analytic sample. Associations for overall cancer, lung cancer, and colon cancer mortality appeared to vary by smoking status, but there was no statistically significant interaction (pinteraction>0.05). No significant interaction with BMI was observed for any of the endpoints (Appendix Table 2).

Table 5.

Association between night shift work and selected endpoints stratified by smoking status, 1988-2010.a

Night shift work duration
Mortality Never 1–5 years 6–14 years ≥15 years p for trend p-interaction
All causes
Never smokers No. of deaths
 HR (95% CI)		 1,863
1.00 (ref)		 1,833
1.04 (0.98, 1.11)		 638
1.20 (1.10, 1.31)		 457
1.04 (0.94, 1.16)		 0.05
Former smokers No. of deaths
 HR (95% CI)		 2724
1.00 (ref)		 2728
0.98 (0.93, 1.03)		 916
1.06 (0.98, 1.14)		 657
1.09 (1.00, 1.19)		 0.01 0.15
Current smokers No. of deaths
 HR (95% CI)		 670
1.00 (ref)		 734
1.07 (0.96, 1.19)		 293
1.08 (0.94, 1.24)		 243
1.21 (1.04, 1.40)		 0.02
All cardiovascular diseases
Never smokers No. of deaths
 HR (95% CI)		 411
1.00 (ref)		 430
1.11 (0.97, 1.27)		 154
1.25 (1.04, 1.51)		 139
1.25 (1.03, 1.52)		 0.01
Former smokers No. of deaths
 HR (95% CI)		 516
1.00 (ref)		 521
1.00 (0.89, 1.13)		 199
1.17 (0.99, 1.38)		 146
1.16 (0.97, 1.40)		 0.03 0.40
Current smokers No. of deaths
 HR (95% CI)		 169
1.00 (ref)		 145
0.86 (0.69, 1.08)		 76
1.10 (0.83, 1.44)		 57
1.11 (0.82, 1.50)		 0.24
Ischemic heart disease
Never smokers No. of deaths
 HR (95% CI)		 138
1.00 (ref)		 143
1.11 (0.88, 1.40)		 53
1.24 (0.90, 1.71)		 55
1.31 (0.96, 1.81)		 0.08
Former smokers No. of deaths
 HR (95% CI)		 192
1.00 (ref)		 175
0.91 (0.74, 1.12)		 71
1.08 (0.82, 1.42)		 62
1.24 (0.93, 1.16)		 0.08 0.26
Current smokers No. of deaths
 HR (95% CI)		 63
1.00 (ref)		 57
0.91 (0.63, 1.31)		 41
1.63 (1.09, 2.43)		 28
1.41 (0.90, 2.22)		 0.02
All cancer
Never smokers No. of deaths
 HR (95% CI)		 668
1.00 (ref)		 672
1.05 (0.95, 1.17)		 205
1.10 (0.94, 1.28)		 1357
0.94 (0.78, 1.13)		 0.77
Former smokers No. of deaths
 HR (95% CI)		 1085
1.00 (ref)		 1102
0.99 (0.91, 1.08)		 339
1.01 (0.90, 1.15)		 235
1.04 (0.90, 1.20)		 0.58 0.18
Current smokers No. of deaths
 HR (95% CI)		 300
1.00 (ref)		 342
1.11 (0.94, 1.29)		 117
1.00 (0.80, 1.24)		 118
1.35 (1.09, 1.68)		 0.03
Lung cancer
Never smokers No. of deaths
 HR (95% CI)		 36
1.00 (ref)		 47
1.31 (0.84, 2.03)		 12
1.27 (0.65, 2.48)		 9
1.09 (0.51, 2.33)		 0.80
Former smokers No. of deaths
 HR (95% CI)		 314
1.00 (ref)		 309
0.99 (0.84, 1.16)		 90
0.91 (0.71, 1.15)		 66
0.96 (0.73, 1.26)		 0.57 0.17
Current smokers No. of deaths
 HR (95% CI)		 140
1.00 (ref)		 149
1.11 (0.86, 1.42)		 60
1.16 (0.84, 1.61)		 67
1.88 (1.36, 2.62)		 <0.001
Colorectal cancer
Never smokers No. of deaths
 HR (95% CI)		 64
1.00 (ref)		 63
1.08 (0.76, 1.53)		 23
1.30 (0.80, 2.12)		 20
1.43 (0.85, 2.39)		 0.13
Former smokers No. of deaths
 HR (95% CI)		 93
1.00 (ref)		 92
0.98 (0.73, 1.32)		 25
0.88 (0.56, 1.38)		 24
1.19 (0.75, 1.89)		 0.60 0.95
Current smokers No. of deaths
 HR (95% CI)		 20
1.00 (ref)		 14
0.74 (0.33, 1.67)		 7
1.07 (0.40, 2.87)		 6
0.94 (0.32, 2.80)		 0.89
Open in a new tab

Boldface indicates statistical significance (p<0.05).

HR, hazard ratio; NHS: Nurses’ Health Study.

a

Cox proportional hazard model adjusted for age (continuous), alcohol consumption (none, 0.1-4.9, 5.0-14.9, ≥15.0 g/d), physical exercise (metabolic equivalent values; quintiles), multivitamin use (yes, no), menopausal status (premenopausal, postmenopausal) and post menopausal hormone use (never, past and current), physical exam in the past 2 years (no, yes for symptoms and yes for screenings), healthy eating score (quintiles), smoking status (never, former, current), pack-years (<10, 10-19, 20-39, ≥40 for former smokers; <25, 25-44, 45-64, ≥65 for current smokers), BMI (kg/m2; <21, 21-22.9, 23-24.9, 25-27.4, 27.5-29.9, 30-34.9, ≥35) ) and husband’s education (less than high school, some high school, high school graduate, college, graduate school, missingness).

p-values for interaction were obtained by adding interaction terms into the models and performing likelihood ratio tests.

Discussion

In this large prospective U.S. cohort of nurses, working rotating night shifts for >5 years was associated with a significant increase in all-cause and all-CVD (particularly IHD) mortality. Working ≥15 years of rotating night shift work was associated with a significant increase in lung cancer mortality. There was also a non-significant increase in mortality due to cancer overall, and several specific cancer sites.

Previous studies of shift work and IHD incidence or mortality varied by design and quality, with inconsistent results.50-53 However, results from two prospective follow-up studies48,54 and a recent meta-analysis2 are in line with the present findings. One cohort study of 504 papermill workers observed a significant increase in IHD risk after 10 years of shift work.30 The other study of the NHS cohort, but with only 4 years of follow-up and many fewer cases (n=292), observed an increase in coronary heart disease (both fatal and non-fatal) incidence among night shift workers after both <5 and ≥5 years of rotating night shift work.48 Moreover, a recent meta-analysis2 suggested that shift work increases the risk of coronary events (including IHD and other CVD, mixed incidence and mortality data); the association remained when restricted to higher-quality studies and was stronger for night shift work, which disrupts circadian rhythms more than other types of shift work. Of the five studies accounting for duration of shift work,18,48,54-56 three reported increased risks after longer durations of shift work,18,48,54 with the exception of two case-control studies.55,56 Together with previous evidence,8,10 the current study provides further support for a link between night shift work, especially of longer durations, and increased IHD events.

Several underlying biological mechanisms make such an association plausible: autonomic nervous system activation, an increased inflammatory state, changes in lipid and glucose metabolism, and resulting changes in the risk for atherosclerosis, all of which have previously been described in night shift workers.57 The finding that women with longer durations of rotating night shift work had a higher BMI and tended to gain more weight since age 18 years may be intermediate conditions subsequent to night shift work, and adjustment for them in multivariable models may underestimate associations. However, analyses stratified by BMI did not alter the results.

The moderately increased lung and colorectal cancer mortality among women with ≥15 years of night shift work is consistent with previous reports on the risk of developing incident lung and colorectal among night shift–working women in the same cohort.58,59 Relatively few studies have examined associations between shift work and the risk of these two cancers. For lung cancer, two studies60,61 reported a higher risk among women in nursing, an occupation with a high prevalence of night shift workers, even after adjustment for smoking. A less reliable study15 found no association between night work and lung cancer risk, but this study was based on registry occupational data, hence possibly underestimating associations, and was not able to account for important confounders or effect modifiers, including smoking. Similarly, the same study found no association between night work and colorectal cancer. Among women with ≥30 years of rotating night shift work, breast cancer mortality was non-significantly increased by 47% in the current study, which is somewhat consistent with a previous report in the same cohort14 describing a significant 36% increase in breast cancer incidence. Current results also suggest potentially increased mortality from renal cancer, and myeloma among women with >5 years of rotating night shift work. The number of deaths was <100 for each of these cancers; therefore, power was likely limited in these analyses, warranting additional studies with larger sample sizes.

This study has a few limitations of note. First, night shift work information was not updated after 1988, potentially leading to exposure misclassification because nurses may have changed their night shift work status during the 22-year follow-up. However, much of follow-up was accrued at midlife or around retirement of these nurses, and more detailed job histories in the NHS2 cohort (a similar cohort, which comprised 116,434 younger women, all registered U.S. nurses) suggest that the percentage of nurses working rotating night shifts declines from roughly 40% in their early 20s, to <5% after age 45 years, with only very few women (<2%) starting night shifts midlife or later (unpublished observation). A further limitation of the exposure assessment is that the timing of exposure in earlier adult life is not known (along with the time between exposure and disease diagnosis). Moreover, in the NHS2 cohort, where night shift work history was assessed in much greater detail, <10% of all nurses worked permanent night shifts. A potential concern is that these nurses may not have classified themselves as working rotating night shifts in the current analysis; yet, exclusion of these women did not change results of a prior analysis.13 Nonetheless, any remaining non-differential misclassification of the exposure assessment could have biased results, albeit likely toward the null in most situations. Another concern relates to multiple comparisons; however, when requiring a p-value of 0.01 for significance, the results remain largely unchanged. Lastly, it is difficult to differentiate the effect of night shift work from lack of sleep; however, sensitivity analyses restricted to women who reported 6–8 sleep hours (not including nap times) per day did not change results materially (data not shown).

This study also has considerable strengths. It is one of the largest prospective cohort studies worldwide with a high proportion of rotating night shift workers and long follow-up time. A single occupation (nursing) provides more internal validity than a range of different occupational groups, where the association between shift work and disease outcomes could be confounded by occupational differences. Shift work is more common in people of lower SES, a risk factor for multiple diseases. Using a cohort of nurses with relatively homogeneous SES minimized potential confounding. Furthermore, husbands’ education was adjusted to account for residual confounding by SES. Although self-reported duration of rotating night shifts cannot be validated, it is likely that results are accurate, because other self-reports were highly accurate in a similar cohort,62 and previous validations of similar questions (e.g., electric blanket use)63 have shown reasonable reproducibility. Moreover, the prospective study design eliminates recall bias. Finally, information on a range of lifestyle factors was prospectively and repeatedly collected and used for multivariable adjustment to minimize potential confounding by these factors.

In summary, this study confirmed associations of rotating night shift work and total and CVD (particularly IHD) mortality, and suggested associations of shift work with multiple cancer sites. These results add to prior evidence of a potentially detrimental effect of rotating night shift work on health and longevity. Future studies are warranted to explore the potential for residual confounding or effect modification by smoking in relation to lung cancer mortality, and to confirm associations of night shift work with colorectal cancer, kidney cancer, and myeloma. To derive practical implications for shift workers and their health, the role of duration and intensity of rotating night shift work and the interplay of shift schedules with individual traits (e.g., chronotype) warrant further exploration.

Supplementary Material

Acknowledgments

This research was supported by National Cancer Institute grants CA50385 and CA87969, NIH Heart, Lung and Blood Institute grant HL034594, and CDC/The National Institute for Occupational Safety and Health grant OH009803. Gu was supported by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics at the National Cancer Institute. We would like to thank the participants and staff of the Nurses’ Health Study cohort for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, and WY. The authors assume full responsibility for analyses and interpretation of these data. We also thank Tricia Li for help with SAS programming.

Footnotes

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 citable 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.

The authors report no conflicts of interest. No financial disclosures were reported by the authors of this paper. The authors alone are responsible for the content and writing of the paper. The sponsors had no role in the design and conduct of the study; the collection, analysis, and interpretation of the data; or the preparation, review, or approval of the manuscript. Researchers of this article are independent of sponsors. This manuscript is an honest, accurate, and transparent account of the study being reported. No important aspects of the study have been omitted. Any discrepancies from the study as planned have been explained.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS640388-supplement.pdf (255.3KB, pdf)

RESOURCES