Prospective Study of Shift Work and Risk of Coronary Heart Disease in Women
Background The purpose of this study was to examine prospectively the relation of shift work to risk of coronary heart disease (CHD) in a cohort of women.
Methods and Results An ongoing prospective cohort of US female nurses, in whom we assessed (in 1988) the total number of years during which they worked rotating night shifts (at least three nights per month in addition to day and evening shifts), included 79 109 women, 42 to 67 years old in 1988, who were free of diagnosed CHD and stroke. Incident CHD was defined as nonfatal myocardial infarction and fatal CHD. During 4 years of follow-up (1988 to 1992), 292 cases of incident CHD (248 nonfatal myocardial infarction and 44 fatal CHD) occurred. The age-adjusted relative risk of CHD was 1.38 (95% CI, 1.08 to 1.76) in women who reported ever doing shift work compared with those who had never done so. The excess risk persisted after adjustment for cigarette smoking and a variety of other cardiovascular risk factors. Compared with women who had never done shift work, the multivariate adjusted relative risks of CHD were 1.21 (95% CI, 0.92 to 1.59) among women reporting less than 6 years and 1.51 (95% CI, 1.12 to 2.03) among those reporting 6 or more years of rotating night shifts.
Conclusions These data are compatible with the possibility that 6 or more years of shift work may increase the risk of CHD in women.
Shift work refers to work patterns that extend beyond the conventional 8-hour work day and that potentially disrupt workers’ normal biological and/or social diurnal rhythms. The varieties of shift work include permanently displaced work hours (including night work), rotating shift work (alternation between thirds of the 24-hour cycle), and unscheduled work hours.1 Of these, rotating night shifts are particularly disruptive for sleep, wakefulness, eating patterns, and social activities.2 Rotating night shifts also are associated with reduced job-related performance and higher levels of perceived stress.3 Additionally, several studies have reported a higher prevalence of coronary risk factors among rotating shift workers, including increased cigarette consumption,4 higher blood pressure,5 and increased serum cholesterol, glucose, and uric acid levels and urinary adrenaline excretion.6
It remains unclear whether rotating night shifts increase the risk of coronary heart disease (CHD). The majority of epidemiological investigations have been cross-sectional7 8 9 or retrospective.10 11 To date, two prospective studies have examined the association between shift work and cardiovascular disease: one study reported no association,12 but the other13 found a relative risk of 1.4 among shift workers compared with day workers.
Data on women are particularly sparse because men constitute the majority of shift workers in the labor force (eg, firefighters, factory workers, taxi drivers, cooks, and security personnel). The major exception to this pattern is female nurses.14 The Nurses’ Health Study afforded a unique opportunity to assess prospectively the association between rotating night shift and CHD risk in women.
The Nurses’ Health Study Cohort
The Nurses’ Health Study cohort was established in 1976, when 121 700 female registered nurses 30 to 55 years old completed a mailed questionnaire requesting information about risk factors for cancer and CHD, including current and past smoking habits and past personal history of myocardial infarction (MI), angina, cancer, diabetes, hypertension, and high serum cholesterol levels. Since then, follow-up questionnaires have been mailed every 2 years to the entire cohort to update information on cardiovascular risk factors and the occurrence of major illnesses. Further details of the Nurses’ Health Study have been described elsewhere.15
In 1988, the entire cohort was asked the question: “What is the total number of years during which you worked rotating night shifts (at least 3 nights per month in addition to days or evenings in that month)?” The responses were categorized as never, 1 to 2, 3 to 5, 6 to 9, 10 to 14, and 15 years or more. In our analyses, we further categorized the cohort into women who had never engaged in rotating night shifts versus those who had ever done so. The questionnaire did not ascertain whether the women were currently engaged in rotating night shifts in 1988. Instead, we ascertained the cohort members’ current work status, including whether they were retired, full-time homemakers, or engaged in part-time or full-time paid employment (either nursing or nonnursing occupations).
Ascertainment of CHD
The study end points comprised incident cases of nonfatal MI and fatal CHD occurring after the return of the 1988 questionnaire but before June 1, 1992. All nurses who reported having a nonfatal MI were asked for permission to review medical records. Cases were confirmed if they met the diagnostic criteria of the World Health Organization (ie, symptoms plus either cardiac enzyme level elevations or diagnostic ECG changes).16 Medical records were reviewed by physicians who were blinded to exposure status. An MI was defined as probable if medical records were not available but hospitalization was required and confirmatory information was obtained by interview or letter. The present analyses included both definite and probable cases. Of the total CHD cases included here, 88% were “definite” by our criteria. The follow-up rate for nonfatal MI through 1992, calculated as a percentage of the total person-years of follow-up, was 98%.
The ascertainment of death included a systematic search of the National Death Index17 to identify deaths among participants who did not respond during each questionnaire cycle. This search was supplemented by reports from next of kin and postal authorities. More than 98% of deaths in the cohort were estimated to have been identified by this method.17 If death appeared to be of vascular causes, written permission was requested from the next of kin (subject to the regulations of vital records offices) to review the medical records. Fatal CHD was defined as fatal MI confirmed by hospital records or at autopsy or as CHD recorded on the death certificate if this was the underlying and most probable cause given and there was previous evidence of CHD. In no instance was the cause on the death certificate accepted without corroboration. Total CHD was defined as nonfatal MI plus fatal CHD.
The primary analysis of follow-up data from 1988 to 1992 used incidence rates with person-years of follow-up as the denominator. Duration of shift work was assessed in 1988. Other variables, such as cigarette smoking and personal history of hypertension, hypercholesterolemia, and diabetes, were updated according to the information provided by the participants on the 1990 questionnaire.
The relative risk was defined as the CHD incidence rate among women who had ever engaged in shift work divided by the corresponding rate among women who had never done shift work. Relative risks were adjusted for age and categorized in 5-year age groups,18 and 95% CIs were calculated.19 We also used proportional-hazards models20 to adjust for multiple risk factors, including smoking status (never, former, and current smoker of 1 to 14, 15 to 24, and 25 or more cigarettes per day); alcohol intake (0.0, 0.01 to 4.9, 5.0 to 14.9, 15.0 to 24.9, and 25.0 or more g/d); body mass index (in quintiles); history of hypertension, diabetes mellitus, and hypercholesterolemia; menopausal status; current use of postmenopausal hormones; past use of oral contraceptives; levels of physical activity (0.0, 0.1 to 7.9, 8.0 to 15.9, 16.0 to 29.9, 30.0 to 49.9, and 50 or more MET-hours per week, where 1 MET-hour is the metabolic rate equivalent to being at rest for 1 hour); quintiles of vitamin E intake; average aspirin use (<1, 1 to 6, and >6 tablets per week); and parental history of MI before age 60.
Of the total cohort of 121 701 women, 11 560 were excluded because they were deceased by 1988 (n=3119) or they had been previously diagnosed with MI or angina (n=7138) or cerebrovascular disease (n=1303). Of the 110 141 eligible women who were free of diagnosed CHD and cerebrovascular disease in 1988, 79 109 women (71.8%) answered the question on shift work. We compared the characteristics of the 79 109 women who completed the shift work question with the 31 032 women who did not. Women who answered the question were less likely to have been current smokers (18.5%) in 1988 than women who did not answer the question (29.7%). In other respects, the nonrespondents did not differ substantially from respondents in terms of their profile of major cardiovascular risk factors, including the prevalence of hypertension (26.7% among respondents compared with 24.9% among nonrespondents), diabetes (3.9% among respondents compared with 4.3% among nonrespondents), and hypercholesterolemia (23.3% among respondents compared with 17.9% among nonrespondents). Importantly, nonrespondents did not appear to be sicker than respondents. The 4-year incidence of CHD among the nonrespondents was very similar to the incidence among the respondents (age-adjusted relative risk of incident CHD among nonrespondents compared with respondents, 1.19; 95% CI, 0.96 to 1.48). We therefore concluded that the exclusion of women who did not respond to the 1988 question on shift work was unlikely to have introduced a major bias into our analyses.
Of the 79 109 women who completed the shift work question in 1988, 32 153 (40.6%) reported that they had never done shift work, whereas 46 956 reported that they had done some shift work. Compared with women who had never done shift work, women who reported ever doing shift work reported a higher age-adjusted prevalence of current smoking (19.2% versus 17.4%) and higher mean body mass index (25.1 versus 24.7). Comparing across women reporting different durations of rotating night shift (Table 1⇓), we found that longer durations of shift work were associated with higher age-adjusted prevalence of current smoking, hypertension, diabetes, body mass index, and levels of physical activity. An inverse relation was also found between mean daily alcohol consumption and duration of rotating night shifts. Compared with women who had done no shift work, a slightly lower proportion of shift workers reported being married in 1980, when marital status was ascertained in the cohort.
We observed a total of 292 incident cases of CHD (248 nonfatal MI and 44 fatal CHD) during 302 964 person-years of follow-up. Having ever worked in rotating night shifts was associated with an ≈1.4-fold increase in the age-adjusted risk of total incident CHD compared with never working shift work (Table 2⇓). Adjusting for a broad variety of cardiovascular risk factors did not substantially alter this estimate.
Increasing durations of shift work were associated with higher relative risks of CHD (Table 3⇓) (probability value for linear trend across durations of shift work, P=.04 including all women and P=.2 after non–shift workers were excluded). We categorized duration of shift work into “less than 6 years” and “6 or more years” on the basis of a priori evidence13 suggesting that the risk of CHD is increased after 6 or more years of rotating night shifts. Compared with women who had never done shift work, the multivariate-adjusted relative risks of total CHD were 1.21 (95% CI, 0.92 to 1.59) among women reporting less than 6 years and 1.51 (95% CI, 1.12 to 2.03) among women reporting 6 or more years of rotating night shifts.
Restricting the analyses to women in the cohort who were in paid employment in 1988 did not substantially affect the relative risk estimates. Between 1988 and 1992, 153 cases of incident CHD occurred among 41 769 women in paid employment (both nursing and nonnursing occupations). Compared with currently employed women who had never done shift work, the multivariate relative risk of total CHD was 1.03 (95% CI, 0.66 to 1.60) among women reporting less than 6 years and 1.70 (95% CI, 1.10 to 2.64) among women reporting 6 or more years of rotating night shifts. When we further restricted the study population to women who were currently employed as nurses (n=34 707), we observed 103 cases of CHD during 4 years of follow-up. The multivariate-adjusted relative risk of total CHD was 1.01 (95% CI, 0.62 to 1.64) among nurses reporting less than 6 years and 1.70 (95% CI, 1.10 to 2.64) among nurses reporting 6 or more years of rotating night shifts.
Although all the subjects in the study were registered nurses at the inception of the study in 1976, our findings could have been explained in part by differences in the socioeconomic backgrounds of women who chose to do shift work. Thus, among women who had never done shift work, 4.5% were married to men who had completed less than a high school education. The corresponding age-adjusted percentages among women reporting various durations of shift work were 4.1% (1 to 2 years of rotating night shift), 4.3% (3 to 5 years), 5.0% (6 to 9 years), 5.8% (10 to 14 years), and 6.8% (15 or more years of shift work). We therefore performed a multivariate analysis adjusting for husband’s educational attainment, in addition to all other cardiovascular risk factors. The five categories of spouse’s educational attainment, which was ascertained on the 1992 questionnaire, were less than a high school education, some high school education, high school graduate, college graduate, and graduate school education. Two hundred sixty-two incident cases of nonfatal MI occurred among 73 400 married women during 4 years of follow-up. Compared with women who had never done shift work, the multivariate-adjusted relative risk of total nonfatal MI was 1.25 (95% CI, 0.94 to 1.68) among women reporting less than 6 years and 1.47 (95% CI, 1.07 to 2.02) among women reporting 6 or more years of rotating night shifts.
Finally, we examined the relation of shift work to total and cause-specific mortality. During 4 years of follow-up, a total of 738 deaths from all causes occurred in the study population. Compared with women who had never done shift work, the age-adjusted relative risks of total mortality were 0.91 (95% CI, 0.79 to 1.05) among women reporting less than 6 years and 1.29 (95% CI, 1.10 to 1.52) among women reporting 6 or more years of rotating night shifts. The excess in risk of total mortality was largely explained by the increased risk of cardiovascular disease mortality among shift workers. A total of 95 deaths from cardiovascular disease (CHD plus cerebrovascular disease) occurred during follow-up. Compared with women who had never done shift work, the age-adjusted relative risks of cardiovascular mortality were 0.91 (95% CI, 0.61 to 1.36) among women reporting less than 6 years and 1.46 (95% CI, 0.95 to 2.23) among women reporting 6 or more years of rotating night shifts. No relation was found between shift work and other major causes of death, including total cancer deaths (n=488) or deaths from accidents and suicide (n=40). Relative to women who had never done shift work, the age-adjusted relative risks of total cancer mortality were 0.88 (95% CI, 0.74 to 1.05) among women reporting less than 6 years and 1.17 (95% CI, 0.96 to 1.43) among women reporting 6 or more years of rotating night shifts.
These data are compatible with the possibility that exposure to 6 or more years of shift work may increase the risk of CHD in women. Previous epidemiological investigations of shift work and cardiovascular disease have been predominantly cross-sectional7 8 9 or retrospective10 11 in design. The two prospective studies of shift work have reported conflicting results. The study by Taylor and Pocock12 used medical records from 10 British companies to calculate the cardiovascular disease mortality rate over a 12-year follow-up period among ≈8700 workers. Workers were categorized into day workers, shift workers, and former shift workers currently in day work. Their mortality rates were then compared with “expected” national age-specific CHD mortality rates to calculate standardized mortality ratios. No statistically significant differences in standardized mortality ratios were found between groups for atherosclerotic heart disease: 94 for day workers, 103 for shift workers, and 125 for former shift workers. This study was subsequently criticized21 for using national mortality rates instead of day workers as the reference group, thereby failing to take account of the healthy worker effect.
The most detailed data on risk of cardiovascular disease by duration of shift work were reported in a 14-year follow-up study of 504 male workers in a Swedish paper mill.13 Although it was based on a small number of events (25 cases of MI and 18 cases of angina), the authors found an overall age-, smoking- and hypertension-adjusted relative risk of 1.4 among shift workers. The relative risk of CHD increased with longer duration of shift work: 1.5 for 2 to 5 years; 2.0 for 6 to 10 years; 2.2 for 11 to 15 years; and 2.8 for 16 to 20 years of shift work (probability value for linear trend, P<.05). These results are generally consistent with our findings in a cohort of female nurses, except that we were able to adjust for a much wider range of cardiovascular risk factors.
Some important limitations of our study are that we did not collect detailed work histories on women in the cohort; nor did we ascertain their current shift work status. Two types of selection processes could bias the association between shift work and cardiovascular disease.1 First, selection into shift work may operate through workplace hiring procedures or through the worker’s estimate of her own ability to withstand irregular working hours. If healthier workers tend to be selected into shift work, this would bias any association with illness toward the null. Alternatively, a spurious positive association with heart disease might arise if workers with less healthy habits (such as workers from lower socioeconomic backgrounds) choose to do shift work. Our study suggests that the latter type of selection process is more likely to occur, although the excess CHD risk associated with shift work persisted after stratification for spouse’s educational status.
The other type of potential bias is selection out of shift work, which has been widely observed in occupational studies.22 If workers drop out of shift work because of illness, this would bias the relative risk downward among current shift workers and increase the risk among former shift workers. However, if dropout from shift work due to illness had been a major factor, we might have expected to observe higher relative risks among women reporting shorter durations of shift work, whereas the opposite was true.
Shift work may be associated with increased risk of CHD for several reasons. Studies have reported a higher prevalence of CHD risk factors among shift workers compared with day workers, including increased cigarette consumption4 and higher blood pressure.5 De Backer et al23 found that workers with the most irregular working hours had higher total and LDL cholesterol and lower HDL. These differences in lipids were independent of smoking habit, obesity, dietary factors, and physical activity. In an experimental study of day workers who were suddenly rescheduled to night work, increased levels of serum cholesterol, glucose, uric acid, and urinary adrenaline excretion were reported.6 These levels fell to normal upon return to day work. If raised blood pressure and serum cholesterol are the mechanisms by which shift work raises the risk of CHD, then controlling for these variables, as we did in our multivariate models, may constitute statistical overadjustment.
The overall magnitude of the association found in our study is modest. Although a large study can rule out chance, an observational study like ours may have involved uncontrollable confounding at least as big as the effects observed. Future studies should investigate the relations between shift work and other putative cardiovascular risk factors, including psychosocial job strain24 and social isolation.25
This study was supported by research grants HL-34594 and CA-40356 from the National Institutes of Health. Dr Manson is supported by a Merck/Society for Epidemiological Research grant award. The authors are indebted to the participants in the Nurses’ Health Study for their continuing cooperation and to Mark Shneyder, Karen Corsano, MA, Gary Chase, Barbara Egan, and Lisa Dunn for their expert help.
- Received April 4, 1995.
- Revision received May 30, 1995.
- Accepted July 24, 1995.
- Copyright © 1995 by American Heart Association
Aanonsen A. Medical problems of shift-work. Ind Med Surg. 1959;28:422-427.
Thiis-Evensen E. Shift work and health. Ind Med Surg. 1958;27:493-497.
Koller M. Health risks related to shift work. Int Arch Environ Health. 1983;53:59-75.
Alfredsson L, Karasek R, Theorell T. Myocardial infarction risk and psychosocial work environment: an analysis of the male Swedish working force. Soc Sci Med. 1982;16:463-467.
Glazner LK. Shift work and its effects on fire fighters and nurses. Occup Health Safety. 1992;61:43-57.
Rose GA, Blackburn H. Cardiovascular Survey Methods. 2nd ed. Geneva, Switzerland: World Health Organization; 1982.
Stampfer MJ, Willett WC, Speizer FE, Dysert DC, Lipnick R, Rosner B. Test of the National Death Index. Am J Epidemiol. 1984;119:837-839.
Rothman KJ, Boice JD Jr. Epidemiologic Analysis With a Programmable Calculator. NIH publication 79-1649. Washington, DC: Public Health Service; 1979.
Miettinen OS. Estimation and estimability in case-referent studies. Am J Epidemiol. 1976;103:226-235.
Cox DR. Regression models and life-tables. J R Stat Soc B. 1972;34:187-220.
Olsen O, Kristensen TS. Impact of work environment on cardiovascular diseases in Denmark. J Epidemiol Community Health. 1991;45:4-10.
Checkoway H, Pearce NE, Crawford-Brown DJ. Research Methods in Occupational Epidemiology. New York, NY: Oxford University Press; 1989.
De Backer G, Kornitzer M, Peters H, Dramaix M. Relationship between work rhythm and coronary risk factors. Eur Heart J. 1984;5(suppl 1):307.
Karasek R, Theorell T. Healthy Work. New York, NY: Basic Books; 1990.
Berkman L, Syme SL. Social networks, host resistance and mortality: a nine-year study of Alameda County residents. Am J Epidemiol. 1976;109:186-204.