Rotating night shift work and risk of multiple sclerosis in the Nurses’ Health Studies

Abstract

Objectives:

Night shift work has been suggested as a possible risk factor for multiple sclerosis (MS). The objective of the present analysis was to prospectively evaluate the association of rotating night shift work history and MS risk in two female cohorts, the Nurseś Health Study (NHS) and NHSII.

Methods:

A total of 83,992 (NHS) and 114,427 (NHSII) women were included in this analysis. We documented 579 (109 in NHS, 470 in NHSII) incident physician confirmed MS cases (moderate and definite diagnosis), including 407 definite MS cases. The history (cumulative years) of rotating night shifts (≥3 nights/month) was assessed at baseline and updated throughout follow-up. Cox proportional hazards models were used to estimate hazard ratios (HR) and 95% confidence intervals (CIs) for the association between rotating night shift work and MS risk adjusting for potential confounders.

Results:

We observed no association between history of rotating night shift work and MS risk in NHS (1–9years: HR 1.03, 95%CI 0.69–1.54; 10+years: 1.15, 0.62–2.15) and NHSII (1–9years: HR 0.90, 95%CI 0.74–1.09; 10+years: 1.03, 0.72–1.49). In NHSII, rotating night shift work history of 20+ years was significantly associated with MS risk, when restricting to definite MS cases (1–9years: HR 0.88, 95%CI 0.70–1.11; 10–19years: 0.98, 0.62–1.55; 20+years: 2.62, 1.06–6.46).

Conclusions:

Overall, we found no association between rotating night shift work history and MS risk in these two large cohorts of nurses. In NHSII, shift work history of 20 or more years was associated with an increased risk of definite MS diagnosis.

INTRODUCTION

Increasing evidence suggests that night shift work and exposure to light at night are associated with a higher risk of chronic disease outcomes including cancer and cardiovascular disease^1–4. Recent studies show that night shift workers may also be at an increased risk of autoimmune diseases such as multiple sclerosis (MS), psoriasis and hypothyroidism^5–9. Two previous studies found that shift work was associated with a higher risk of MS, particularly among participants exposed early in life (before age 20 years)^{6 10}. A third study found no overall association between shift work and risk of MS⁵. The epidemiological evidence is still very limited and no prospective study has evaluated the possible link between night shift work and MS risk.

The association between night shift work and MS risk is biologically plausible. Night shift work causes circadian disruption, sleep deprivation and melatonin suppression that may in turn impact the immune function¹¹. In mouse models of MS, circadian rhythms such as liver clock gene expression, corticosterone and leptin rhythms, have been shown to be altered, which is supportive of a link between circadian disruption and autoimmune inflammatory disease¹². In experimental and observational human studies circadian disruption and sleep deprivation, two common consequences of night shift work, have been linked with increased inflammatory biomarkers and lower melatonin levels^13–16. Melatonin is a well-known immunomodulator and may be a potential key hormone in MS onset¹⁷. Melatonin levels have been shown to correlate negatively with multiple sclerosis activity in humans¹⁸. In clinical studies circulating melatonin levels were found to be predictive of MS severity¹⁹ and exogenous melatonin was shown to reduce inflammatory response in MS patients²⁰. These findings suggest that night work that involves circadian disruption may have implications for MS etiology and progression.

In the present study we prospectively examined the association between the history (cumulative years) of rotating night shift work and occurrence of MS in two prospective cohorts of US women, the Nurses’ Health Study (NHS) and NHSII.

METHODS

Study participants

The NHS recruited 121,701 female nurses aged 30–55 years old in 1976 and the NHSII 116,430 female nurses aged 25–42 years old in 1989. Every two years, women completed a questionnaire to update lifestyle exposures and medical conditions. The first assessment of night work history was in 1988 in NHS and in 1989 in NHSII; the return dates of these questionnaires were the start of follow-up (baseline) for each cohort, respectively. After excluding participants with missing shift work information at baseline, MS diagnosis before baseline or missing date of MS diagnosis, there were 83,992 women in NHS and 114,427 women in NHSII included in the analysis.

Standard protocol approvals, registrations, and patient consents

This study was approved by the institutional review board of Brigham and Women’s Hospital Investigators received written informed patient consent to perform this study.

Ascertainment of MS cases

Participants with newly diagnosed MS self-reported on biennial questionnaires were asked for permission by investigators to contact their neurologists and review their medical records, clinical history (including diagnosis date and date of first symptoms), and the neurologist’s certainty of diagnosis (definite, probable, possible, not MS). Since 2003, the study neurologist (T.C.) has reviewed all medical records. If a neurologist did not respond, a questionnaire was sent to the participant’s internist. For 90% of women with MS, the treating physician was a neurologist. In most cases, diagnosis was supported by magnetic resonance imaging (MRI) findings (>70% NHS; >90% NHSII)²¹. For these analyses, we defined confirmed cases as those with definite or probable MS according to neurologist or internist. Using this definition, we confirmed a total of 579 new MS cases in both cohorts; 109 of them occurred in the NHS cohort and 470 cases in NHSII. Secondary analyses were restricted to the 407 definite cases of MS (NHS: 73 cases, NHSII: 334 cases).

Night-shift work assessment

Information on total years during which the nurse had worked rotating night shifts was available at baseline (1988/1989 questionnaires) in both cohorts. Specifically, the question was: “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)?” The pre-specified response categories were: “Never; 1–2 years; 3–5 yrs; 6–9 yrs; 10–14 yrs; 15–19 yrs; 20 years or more”. In the NHSII this information was updated over follow-up by asking women how many months of rotating night shifts they worked during the two-year periods subsequent to each biennial questionnaire in 1991, 1993, 1997, 2001, 2005, and 2007; pre-specified response categories were: none, 1–4 months, 5–9, 10–14, 15–19, and ≥20 months. Because the NHSII questionnaires of 1995 and 2003 did not contain a question updating this information, retrospective assessment of rotating night shifts for the periods 1997–1999 and 2001–2003 were included on the 2001 and 2005 questionnaires, respectively. We calculated the total number of years having worked rotating night shifts for at least 3 times per month over a nurse’s career. If shift work information was missing in one of the follow-up cycles, we assigned 0 months to that cycle and kept women in the analysis.

Covariates

Information on demographics, lifestyle and diet was reported on biennial questionnaires and updated through follow-up. Women reported information on state of residence at ages 15 (1992 in NHS; 1993 in NHSII) and ancestry (1992 in NHS; 1989 in NHSII; South European, Scandinavian, other Caucasian, African American, Hispanic, Asian, Native American (NHS only), other) and were categorized as previously described ²¹. Women updated smoking status every two years; pack-years of smoking were derived from this information. Women completed comprehensive validated semi-quantitative food frequency questionnaires every four years and supplemental vitamin D intake (IU/day) was derived using reported intake²². Women reported their current height (1976 in NHS; 1989 in NHSII) and weight at age 18 (1980 in NHS; 1989 in NHSII); body mass index (BMI) at age 18 was calculated by dividing age 18 weight (in kilograms) by reported height (in meters) squared.

Statistical analysis

Baseline characteristics (NHS: 1988; NHSII: 1989) of the study population were age-standardized and reported by rotating night shift work history and cohort. Participants contributed person-time to the follow up period from the date of first shift work assessment (1988 in NHS and 1989 in NHSII) to the date of diagnosis of MS, death from any cause, or end of follow-up (May 31, 2004 for NHS and May 31, 2011 for NHSII). Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) related to MS risk (moderate and definite diagnosis), for each shift work history category. The reference group comprised of nurses who never worked rotating night shifts. Pre-determined cut points of shift work history (total years) were used to classify participants (0, 1–9 years, 10+ years of rotating night shift work) in both cohorts. An additional exposure classification was used only in NHSII, where power was sufficient, to capture even longer durations of cumulative updated shift work history (0, 1–9, 10–19, 20+ years of rotating night shift work). To account for potential confounding by age and time all models were adjusted for age (5-year age groups) and 2-year time intervals. The fully (multivariable) adjusted models were additionally adjusted for known and suspected MS risk factors^{23 24} such as ancestry (southern European, Scandinavian, other Caucasian, and non-white), latitude of state of residence at age 15 (northern, middle, southern states), BMI at age 18 (<18.5, 18.5–20.9, 21–22.9, 23–24.9, 25–26.9, 27–29.9, ≥30 kg/m²), intake of vitamin D from supplements (none, <400, ≥400 IUs/day), pack years of smoking (0, <10, 10–24, 25+ pks/yr). Missing indicators were used for covariates with missing information. Tests for trend were conducted by modeling the median values of each category of night shift work history as a continuous variable. In secondary analysis we used a more stringent definition for MS by restricting the analysis to only definite diagnosis of MS and recomputed HRs and 95%CI according to cumulative years of rotating night shift work. Stratified analyses were performed to examine effect modification by BMI at age 18 and smoking. In sensitivity analyses, the date of first symptoms of MS (instead of date of MS diagnosis) was used for person-time calculation and time to event analysis. In these analyses, subjects with reported symptoms before baseline were additionally excluded in order to restrict to truly incident MS cases and minimize the potential for reverse causality. Risk estimates (HRs and 95% CI) were recalculated for all exposure categories and both MS definitions (moderate/definite and definite diagnoses). All p values were 2-sided. All analyses were conducted using Statistical Analysis System (SAS), version 9.3.

RESULTS

As expected, compared to women in NHSII, women in NHS were older at baseline (54.9 yrs; SD 7.2 vs 34.3 yrs; SD 4.7) and at age of MS diagnosis (59.2 years; SD 6.9 vs 44.5 years; SD 6.8). In both cohorts, women with longer history of shift work at baseline were older, had greater BMIs and were more likely to be non-white and heavy smokers, compared to never shift workers (Table 1). Overall, we observed no significant association between number of years of rotating night shifts at baseline and MS risk among women in the NHS cohort (1–9 yrs: multivariable adjusted HR 1.03, 95%CI 0.69, 1.54; 10+ yrs: HR 1.15 95%CI 0.62, 2.15; P_Trend=0.61), (Table 2). Similarly, in NHSII, neither baseline nor updated history of rotating night shift work was associated with MS risk. In NHSII, nurses who had worked rotating night shifts for more than 20 years had an 80% increased MS risk in the fully adjusted models, compared to never shift workers, but differences were not statistically significant. In secondary analyses restricted to definite MS cases, associations between longer durations of rotating night shift work and MS risk became stronger in both cohorts (Table 3). Using this more stringent definition of MS, 20+ years of rotating night shift work was associated with a 2.6-fold increase in MS risk which remained significant in the fully adjusted models (HR=2.61, 95%CI 1.06–6.47; P_Trend =0.50), although this was based on 5 MS cases. Further, in stratified analyses we found no evidence of effect modification by BMI at age 18 and smoking, but estimates were unstable due to small case numbers (results not shown).

Table 1.

Age-standardized^* baseline characteristics of the study population by rotating night shift work history (NHS I: 1988, NHS II: 1989)

	NHS I (N= 83,992)			NHS II (N=114,427)
	Shift work history			Shift work history
	Never (N=33,882)	1–9 years (N=39,968)	10+ years (N=10,142)	Never (N=43,487)	1–9 years (N=65,697)	10+ years (N=5,243)
Age (years); mean (SD)	54.5(7.1)	54.8(7.1)	56.6(6.9)	34.3(4.7)	34.2(4.7)	36.9(3.4)
BMI age 18 (kg/m2); mean (SD)	21.3(2.9)	21.4(2.9)	21.8(3.3)	21.1(3.2)	21.3(3.4)	22.0(4.2)
Pack-years of smoking,(%)
0 pack-yrs	46.2	44.0	42.7	67.3	64.5	60.1
1–9 pack-yrs	16.9	17.9	14.6	16.1	17.5	18.6
10–24 pack-yrs	15.5	15.7	15.8	14.3	15.4	18.0
25+ pack-yrs	21.5	22.4	27.0	2.4	2.6	3.3
Vitamin D (IUs /day) supplement use (%)
None,	59.7	59.0	60.0	58.9	57.9	53.4
<400 IU/d,	17.7	18.6	18.9	19.2	20.0	21.9
>= 400 IU/d,	22.6	22.4	21.1	22.0	22.2	24.7
Latitude at age 15 (%)
Northern states	42.9	41.8	38.6	34.8	34.1	31.8
Middle states	48.2	50.5	53.8	48.2	51.5	57.3
Southern states	8.9	7.7	7.6	17.0	14.4	10.9
Ethnicity (%)
Southern European	14.4	14.5	15.2	13.4	13.8	14.7
Scandinavian	4.2	4.2	3.5	4.3	4.5	3.8
other Caucasian	60.5	61.5	54.5	73.3	71.6	69.8
non-white	20.9	19.8	26.8	9.1	10.2	11.7

^*Values are standardized to the age distribution of the study population.

Table 2.

Relative risk of multiple sclerosis (probable or definite diagnoses)according to cumulative years of rotating night shift work among women in the Nurses’ Health Study (NHS) (N=109 cases) and NHS II (N=470 cases).

Rotating night shift work history (cumulative years)	Cases/person-years	Age-adjusted HR (95% CI)1	Multivariable-adjusted HR (95% CI) 2	P trend3
NHS I (baseline 10+)
Never	44/507,650	1.00	1.00
1–9 years	52/598,084	1.04 (0.70, 1.56)	1.03 (0.69, 1.54)
10+ years	13/148,033	1.23 (0.66, 2.28)	1.15 (0.62, 2.15)	0.61
NHS II (baseline 10+)
Never	189/870,039	1.00	1.00
1–9 years	254/1,306,269	0.90 (0.74, 1.08)	0.89 (0.74, 1.07)
10+ years	27/103,293	1.18 (0.79, 1.78)	1.11 (0.74, 1.67)	0.57
NHS II (updated 10+)
Never	153/705,558	1.00	1.00
1–9 years	281/1,421,902	0.92 (0.76, 1.12)	0.90 (0.74, 1.09)
10+ years	36/152,141	1.11 (0.77, 1.60)	1.03 (0.72, 1.49)	0.95
NHS II (updated 20+)
Never	153/705,558	1.00	1.00
1–9 years	281/1,421,902	0.92 (0.76, 1.12)	0.90 (0.74, 1.09)
10–19 years	31/140,081	1.04 (0.70, 1.53)	0.97 (0.66, 1.43)
20+ years	5/12,060	2.03 (0.83, 4.95)	1.80 (0.73, 4.40)	0.80

¹Hazard Ratio (HR) and 95% confidence interval (CI) adjusted for age (5 year categories)

²Hazard Ratio (HR) and 95% confidence interval (CI) additionally adjusted for intake of vitamin D supplements (none, <400, ≥400 IUs/day), latitude at age 15 (northern, middle, southern states), pack years of smoking (0, <10, 10–24, ≥25), body mass index at age 18 (<18.5, 18.5–20.9, 21–22.9, 23–24.9, 25–26.9, 27–29.9, ≥30) and ancestry (southern European, Scandinavian, other Caucasian, and non-white)

³Calculated using medians of shift work duration categories fully adjusted for all variables listed in model 2.

Table 3.

Relative risk of multiple sclerosis (definite diagnoses) according to cumulative years of rotating night shift work among women in the Nurses’ Health Study (NHS) (N=73 cases) and NHS II (N=334 cases).

Rotating night shift work history (cumulative years)	Cases/person-years	Age-adjusted HR (95% CI)11	Multivariable-adjusted HR (95% CI)12	P trend3
NHS I (baseline 10+)
Never	28/507,650	1.00	1.00
1–9 years	36/598,084	1.13 (0.69, 1.86)	1.11 (0.68, 1.82)
10+ years	9/148,033	1.35 (0.64, 2.88)	1.27 (0.60, 2.72)	0.47
NHS II (baseline 10+)
Never	135/870,090	1.00	1.00
1–9 years	176/1,306,345	0.87 (0.69, 1.09)	0.86 (0.69, 1.08)
10+ years	23/103,296	1.41 (0.90, 2.20)	1.32 (0.84, 2.07)	0.20
NHS II (updated 10+)
Never	110/705,599	1.00	1.00
1–9 years	197/1,421,986	0.91 (0.72, 1.15)	0.88 (0.70, 1.11)
10+ years	27/152,146	1.19 (0.78, 1.81)	1.10 (0.72, 1.69)	0.96
NHS II (updated 20+)
Never	110/705,599	1.00	1.00
1–9 years	197/1,421,986	0.91 (0.72, 1.15)	0.88 (0.70, 1.11)
10–19 years	22/140,086	1.05 (0.66, 1.66)	0.98 (0.62, 1.55)
20+ years	5/12,060	2.98 (1.21, 7.33)	2.62 (1.06, 6.47)	0.50

¹Hazard Ratio (HR) and 95% confidence interval (CI) adjusted for age (5 year categories)

²Hazard Ratio (HR) and 95% confidence interval (CI) additionally adjusted for intake of vitamin D supplements (none, <400, ≥400 IUs/day), latitude at age 15 (northern, middle, southern states), pack years of smoking (0, <10, 10–24, ≥25), body mass index at age 18 (<18.5, 18.5–20.9, 21–22.9, 23–24.9, 25–26.9, 27–29.9, ≥30); and ancestry (southern European, Scandinavian, other Caucasian, and non-white)

³Calculated using medians of shift work duration categories fully adjusted for all variables listed in model 2.

In sensitivity analyses based on the date of first MS symptoms and additional exclusion of women that reported symptoms before baseline, the direction of the association between night shift work history and MS risk was similar to the main analysis in NHSII and slightly stronger in the NHS cohort, though none of the differences were statistically significant, since these analyses were based on a smaller number of cases (Supplemental Tables 1 and 2).

DISCUSSION

In these two large cohorts of nurses we found no evidence of an association between rotating night shift work and MS risk. In NHS and to a less extent in NHSII, we were only able to evaluate later life onset of MS, due to the older age of the nurses at baseline. In NHSII, MS risk tended to increase with longer history of shift work (20+ years) suggesting that long-term or early career circadian disruption might be critical for MS, but our estimates were based on small case numbers and therefore have to be viewed with caution.

Our study consisted of two large cohorts and for the first time prospectively assessed rotating night shift work in relation to MS risk (diagnosis and first symptoms) and could not confirm an overall positive association as was described in some of the previous studies. In a large population based case-control study in Sweden, including incident (N=1,343) and prevalent (N=5,129) MS cases¹⁰, shift work before age 20 years - but not after 20 years - increased the risk of MS compared to day shift work. ¹⁰ In a subsequent analysis (N=2,337 incident cases) the investigators confirmed the increased MS risk with ever shift work, independently of the age of first exposure, although results were again stronger for early life exposure (<20 yrs) to shift work⁶. A third case-control study (N=1,723 cases) did not observe a significant association between MS and shift work (ever/never) or shift work duration (years)⁵. This study described a small increase in risk with shift work intensity (nights/month), but used blood donors as controls which is generally a healthier group compared to the general population^{25 26}.

There are a few reasons why our findings may be different from previous studies. First, all previous studies, although large in size, were potentially affected by selection and information bias - especially recall of past exposures – which may partly explain their positive results. Second, in our analysis because women were relatively older at baseline (especially in NHS and to a lesser extent in NHSII) and only incident MS cases were included, our findings may only apply to later onset MS, as opposed to previous studies that had no restriction regarding age of MS diagnosis. Last, previous studies included shift workers of both sexes from a variety of occupational sectors (shift work systems including both rotating and permanent night work), whereas ours is a study of female nurses, which is a more homogeneous group in terms of exposure (rotating night shift work with ≥3 nights/month) and confounders.

Although we found no overall association between rotating shift work history and MS, our study suggested a possible increased risk of MS with extreme durations (20+ years) of shift work which was statistically significant when the analysis was restricted to definite diagnosis of MS. This finding suggests that long-term circadian disruption may play a role in MS development and given the lack of a significant linear trend this may represent a threshold effect. Alternatively, women with the longest shift work histories (20+ years) might have started shift work at a younger age, which supports the hypothesis that early life circadian disruption may be most critical for MS development, as suggested in previous studies^{5 10}. It is likely that nurses’ exposure to night shift work began early on in their careers during training and first years of practice. However in our cohorts, exact information on shift work timing such as age of first night shift work was not available. Thus, we could not disentangle the effects of early timing of night shift work to those of long shift work duration.

Long-term night shift work leads to chronic circadian disruption and sleep deprivation that in turn may lead to an altered immune response, according to experimental animal and human studies^{11 12 27}. Suggested underlying pathophysiologic mechanisms include oxidative stress and neuroendocrine changes such as increase in glycocorticoids, pro-inflammatory cytokines and leukocyte count that lead to a chronic inflammatory condition²⁸. In addition, melatonin - a hormone that is typically suppressed by light in night shift workers - has well-known anti-inflammatory, anti-oxidant and anti-apoptotic effects that may well play a role in MS development^{19 29–32}. It has been hypothesized that lower melatonin levels might be predictive of a higher MS risk and worse MS prognosis^{18 19 30}. Our study adds information on the duration of rotating night shift work, suggesting that long-term (20+ years) exposure to frequent (≥3 nights/month) night shift work might play a role in the pathogenesis of MS.

The main strengths of this study include the prospective cohort design, large sample size, cumulative exposure (years) information and the MS case confirmation by a study physician/neurologist. This study also has some limitations. First, in the NHS, rotating night shift work exposure was assessed in 1988 only and information was not updated through follow-up. However, women at baseline were middle-aged and therefore were likely to have correctly reported their cumulative shift work exposure throughout a large part of their careers. Second, although this was a prospective analysis, rotating night shift work history was assessed retrospectively at baseline and exposure misclassification might have occurred, but most likely non-differential (to the null) since all subjects were healthy at baseline. Third, although our analysis was adjusted for most of the known predictors of MS^{23 24}, multivariable models were not adjusted for past history of infectious mononucleosis, since this information was not routinely collected in these cohorts. Infectious mononucleosis is a risk factor for MS that might have temporarily affected the nurses’ work status but most likely not the total years of night shift work, therefore there is a small confounding potential that could have attenuated our findings. Fourth, risk estimates in NHS were unstable since these analyses were based on a small number of incident MS cases. Last, due to the older age of women at baseline in both cohorts –but especially in the NHS cohort we were only able to assess later life MS diagnosis. Given that onset of MS typically occurs in young adulthood, and that early life MS onset may not fully share the same etiology with later in life MS onset, our results need to be interpreted with caution.

In conclusion, in these two large prospective studies we found no overall evidence of an association between rotating night shift work and MS risk. In NHSII, shift work history of 20 or more years was associated with an increased MS risk, based on a small number of MS cases. More studies are needed to confirm the possible role of night shift work in MS development. While larger cohort studies might not be feasible, case-control or register based studies specifically for MS may help delineate this association. In addition to night work other sources of circadian and sleep disruption (e.g. light at night exposure and screen use among adolescents and young adults) might be of relevance for MS development and could be considered in future studies.

Key messages.

What is already known about this subject?

• Shift work that involves circadian disruption has been associated with a wide range of chronic disease outcomes such as cancer, cardiovascular disease and recently autoimmune diseases.

• Night shift work has been suggested as a possible risk factor for multiple sclerosis (MS) but epidemiological evidence is still limited.

What are the new findings?

• This is the first prospective evaluation of night work in relation to MS risk.

• Although we found no overall evidence of an association between night work and MS, risk of a definite MS diagnosis significantly increased after 20 years of night work.

How might this impact on policy or clinical practice in the foreseeable future?

• More studies are needed to confirm the possible role of night shift work in MS development.

• Night workers may benefit from preventive measures and disease screening programs at the workplace, already at early career stages.