Abstract
A growing number of studies have examined associations between night shift work and the risks of common cancers among women, with varying conclusions. We did a meta-analysis to identify whether long-term night shift work increased the risks of common cancers in women. We enrolled 61 articles involving 114,628 cases and 3,909,152 participants from Europe, North America, Asia, and Australia. Risk estimates were performed with a random-effect model or a fixed-effect model. Subgroup analyses and meta-regression analyses about breast cancer were conducted to explore possible sources of heterogeneity. In addition, we carried out a dose–response analysis to quantitatively estimate the accumulative effect of night shift work on the risk of breast cancer. A positive relationship was revealed between long-term night shift work and the risks of breast [OR = 1.316; 95% confidence interval (CI), 1.196–1.448], digestive system (OR = 1.177; 95% CI, 1.065–1.301), and skin cancer (OR = 1.408; 95% CI, 1.024–1.934). For every 5 years of night shift work, the risk of breast cancer in women was increased by 3.3% (OR = 1.033; 95% CI, 1.012–1.056). Concerning the group of nurses, long-term night shift work presented potential carcinogenic effect in breast cancer (OR = 1.577; 95% CI, 1.235–2.014), digestive system cancer (OR = 1.350; 95% CI, 1.030–1.770), and lung cancer (OR = 1.280; 95% CI, 1.070–1.531). This systematic review confirmed the positive association between night shift work and the risks of several common cancers in women. We identified that cancer risk of women increased with accumulating years of night shift work, which might help establish and implement effective measures to protect female night shifters. Cancer Epidemiol Biomarkers Prev; 27(1); 25–40. ©2018 AACR.
This article is featured in Highlights of This Issue,
Introduction
In modern society, the fast-growing productivity demands for working across time zones and night shift work is increasingly prevalent in different industries such as food production, entertainment, health care, and transportation (1). Circadian disruption from electric lighting poses huge challenges on public health, including cardiovascular diseases, neuropsychiatric and endocrine system disorders, and even cancers, in particular breast cancer (2–4). Cancer incidence in industrialized regions is noticeably higher than that in developing countries, suggesting that environmental factors of modern society play an role in cancer etiology. In 2007, the International Agency for Research on Cancer (IARC) has identified “shift work that involves circadian disruption” to be probably carcinogenic (Group 2A), based on “limited evidence in humans for the carcinogenicity of shift work that involves night work,” and “sufficient evidence in experimental animals for the carcinogenicity of light during the daily dark period (biological night)” (5). Among night shift workers, female employees account for a large proportion, and the number of women in nursing is always overwhelming more than male. During 2004–2005, night shift workers accounted for 12.4% of the female working population and 17.4% for European countries (6). Long-term night shift work serves as a potential risk factor for the common cancers in female population.
Much of the research has examined that nocturnal melatonin suppression and circadian rhythm disruption caused by night shift work function as carcinogens that increase tumor incidence (5). Melatonin, primarily produced by pineal gland, was reported to play an important role in inhibiting tumor growth through antioxidation, antiangiogenesis, and regulation of immunity (7). Unnatural light at night reduced melatonin release, which contributed to tumor development. However, melatonin suppression had a negative feedback effect on hypothalamic-pituitary-gonadal axis, promoting gonadotropins secretion (8). Previous studies stated that night shift work increased the risk of hormone-dependent cancers including prostate cancer in men, breast cancer and ovarian cancer in women (9). In addition, another mechanism was related to clock genes expression which played an important role in several cellular processes such as DNA repair and cell apoptosis. Circadian rhythm disturbance among shift workers resulted in changes of clock genes expression, ultimately increased cancer susceptibility (10, 11).
Animal models focused on light at night to study the effect of circadian disruption on cancer incidence (12), whereas epidemiological evidence is limited with differing results. A growing number of females are being exposed to night shift work and employments in different working fields vary in exposure status such as frequency and period time. Nursing group are generally exposed to long-term and high-intensity night shift work. The association between night shift work and cancer risk of female nurses was not covered by published reviews. Breast cancer is the most common cancer in women worldwide. Most of previous meta-analyses emphasized, in particular, the association between night shift work and breast cancer and had controversial conclusions (13–17). Some of them concluded that night shift work was significantly associated with higher risk of breast cancer (13, 14, 17) whereas another study provided limited evidence (16) and also one study reported a small nonsignificant effect of long-term night shift work (15). Thus it still remains unknown whether night shift work elevates the risk of common cancers such as breast cancer, ovarian cancer and lung cancer. Here we performed a meta-analysis to demonstrate the effect of night shift work on the risk of common cancers among women. Meanwhile, the group of female nurses was evaluated for the separate risk estimation of multiple common cancers. By systematically integrating a multitude of previous data, we expected to arrive at a convincing conclusion which would help propose health protection programs for long-term female night shift workers.
Materials and Methods
Literature search
We conducted the meta-analysis following the quality standards of a meta-analysis. An extensive systematic literature search updated to October, 2016 was performed. We searched the keywords “night shift” “shift work” “shift-work” “cancer risk” “cancer mortality” in PubMed, Embase, Medline, and Web of Science databases. Only English articles were enrolled, and no other limitations were restricted. Also, we manually searched citing and reference lists to identify other relevant studies.
Inclusion and exclusion criteria
Studies were identified according to inclusion and exclusion criteria. Studies were included if they met the following criteria: (i) cohort study, case–control study, or nested case–control study within cohort study; (ii) study evaluating cancer risk among women that were ≥18 years old and were exposed to night shift work; (iii) study involving OR, RR, HR, or standardized incidence ratio (SIR) with 95% confidence intervals (CIs) or providing sufficient data to calculate the above parameters. Studies were excluded if they met the following criteria: (i) study involving female cancer risks that could not be separated from that of male; (ii) study providing overlapping or insufficient data; (iii) study involving recurrent cancer.
Data extraction
Extracted information from enrolled studies included first author, published year, number of cases and subjects, OR and corresponding 95% CI, study design, quality score, region, type of cancer, range of night shift work, occupation, variables of adjustment, and exposure assessment. We adopted the shortest and longest exposure time for preceding analysis. Data extraction was performed independently by two investigators and a third author resolved the differences by face-to-face discussion.
Quality evaluation
We used the Newcastle-Ottawa Quality Assessment Scale (NOS; ref. 18) for quality evaluation of eligible studies. NOS adopted the star system with a maximum of nine stars scoring from 0 to 9, which was divided into four parts: participant selection, comparability of study group, exposure assessment, outcome evaluation, and scoring <7 indicated a low quality. NOS quality evaluation was conducted by two investigators independently and a third author settled all disagreements.
Statistical analysis
We evaluated the association between night shift work and female cancer risk using statistical software STATA Version 11.0 (StataCorp). ORs with their corresponding 95% CIs were used as effect measure. Statistical heterogeneity was evaluated by Q and I2 statistics. P < 0.10 and I2 > 50% indicated an existence of statistical heterogeneity and a random-effect model was then carried out, otherwise a fixed-effect model was used (19). To explore the possible sources of heterogeneity regarding breast cancer, meta-regression analyses were performed. Moreover, we conducted subgroup analyses stratified by region, study design, occupation, exposure assessment, number of variables, and quality score.
For dose–response meta-analysis, we retrieved studies which involved at least three levels of exposure categories and information of cases, number of total subjects, person-year, years of exposure in each category were extracted. The midpoint of lower and high boundary was used as average time of night shift exposure. The range of highest category was supposed to be the same as the adjacent category if the upper boundary was not provided (20). Two-stage random-effect model was adopted to estimate the overall dose–response trend.
Begg funnel plot was performed to evaluate the publication bias of enrolled studies and P < 0.05 suggested the evidence of publication bias (15). In the Begg funnel plot, the standard error of logarithm (Log) for OR was plotted against its OR, and Log OR was plotted versus standard error of Log OR for each enrolled study (21).
Results
Literature search and selection of studies
The initial search yielded 368 articles, and 98 articles were retrieved after checking titles and abstracts. Then, we reviewed full texts of these articles and 56 were included according to eligibility criteria. Also, 5 relevant studies were identified by manually searching citing and reference lists. Finally, 61 articles (22–82) were eligible for a comprehensive analysis. The selection flowchart is shown in Fig. 1.
Flow chart of eligible studies selection process. An extensive systematic literature search updated to October, 2016 was performed in PubMed, Embase, Medline, and Web of Science databases, yielding 368 articles, and finally 61 articles were included according to eligibility criteria.
Flow chart of eligible studies selection process. An extensive systematic literature search updated to October, 2016 was performed in PubMed, Embase, Medline, and Web of Science databases, yielding 368 articles, and finally 61 articles were included according to eligibility criteria.
Study characteristics
Sixty-one articles were enrolled in the meta-analysis including 26 cohort studies, 24 case–control studies, and 11 nested case–control studies. Several articles investigated whether the carcinogenic effect of night shift was related to breast cancer estrogen-receptor status (ER+ and ER− breast cancer; ref. 38) or menopausal status (premenopausal and postmenopausal; ref. 46) or different years in employment (60). In addition, several articles reported the risks for a number of cancers including breast cancer, cervical cancer, and colon cancer, and so on (57, 61, 62, 71, 77). One study (82) evaluated associations between night shift work and the risk of breast cancer in two cohorts, respectively [Nurses' Health Study (NHS) and Nurses' Health Study II (NHS2)]. Thus, 67 studies from 61 articles comprising 114,628 cases and 3,909,152 participants were analyzed for the association between night shift work and common cancers at 11 sites of women including breast cancer, digestive system cancer, skin cancer, reproductive system cancer, hematologic system cancer, endocrine system cancer, nervous system cancer, urinary system cancer, lung cancer, bone and soft tissue cancer. Most studies reached a standard of high follow-up rate or favorable response rate. Seventeen out of 67 studies were conducted among nurses, and 49 out of 67 studies evaluated the association between night shift work and the risk of breast cancer. Questionnaire, interview, and databases were adopted for exposure assessment, and we extracted the shortest versus longest history of night shift work as exposure indicator. Forty-nine studies were adjusted for >3 confounders and 18 studies for ≤3 confounders. The main characteristics of the included studies were summarized in Table 1.
Main characteristics of included studies on the relationship between night shift work and the risks of common cancers in female populations
| Study | Published year | No. of cases/No. of subjects | OR (95% CI) | Study design | Quality score | Region | Type of cancer | Range of night shift work | Occupation | Variables of adjustment | Exposure assessment method |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Knutsson A | 2012 | 94/4,036 | 2.02 (1.03–3.95) | Cohort study | 7 | Sweden | Breast cancer | Day vs. night shift | NA | Number of children, alcohol consumption, BMI, height, weight, waist, hip circumference, educational level, smoking menopausal status, status of oral contraceptive use, and hormones other than contraceptives | Questionnaire |
| Carter BD | 2014 | 1,289/16,1004 | 1.27 (1.03–1.56) | Cohort study | 8 | USA | Ovarian cancer | Day vs. rotating shifts | NA | Oral contraceptive use, age at menarche and menopause, tubal ligation, parity, postmenopausal estrogen use, race, family history of breast/ovarian cancers, exercise, BMI, and height | Questionnaire |
| Poole EM | 2011 | 718/181,548 | 0.8 (0.51–1.23) | Cohort study | 7 | USA | Ovarian cancer | Never vs. 20+ years rotating shift work | Nurses | Age, duration of oral contraceptive use, parity, BMI, smoking status, tubal ligation history, menopausal status, family history of ovarian cancer, duration of breas tfeeding, and cohort | Questionnaire |
| Viswanathan AN | 2007 | 515/53,487 | 1.47 (1.03–2.1) | Cohort study | 7 | USA | Endometrial cancer | Never vs. 20+ years rotating shift work | Nurses | Age, age at menarche, age at menopause, parity, BMI duration of oral contraceptive use, use and duration of postmenopausal hormones, hypertension, diabetes, pack-years of smoking | Questionnaire |
| Akerstedt T | 2015 | 463/1,3656 | 1.77 (1.03–3.04) | Cohort study | 8 | Sweden | Breast cancer | Never vs. 21–45 years | NA | Age, education level, tobacco consumption, BMI, having children, coffee consumption, previous cancer, use of hormones including oral contraceptives | Telephone interview |
| Koppes LLJ | 2014 | 2,531/285,723 | 0.87 (0.72–1.05) | Cohort study | 8 | Netherland | Breast cancer | Never vs. regular night work | NA | Age, origin, children in household education, occupation, job tenure (years) | Personal interview |
| Natti J | 2012 | 48/1,649 | 2.82 (1.196–6.645) | Cohort study | 8 | Finland | Unclassified cancer | Day vs. weekly night shift | NA | Age, and smoking status, demographics and health- and work-related factors | Face-to-face interview |
| Schernhammer ES | 2006 | 1,352/115,022 | 1.79 (1.06–3.01) | Cohort study | 7 | USA | Breast cancer | Never vs. 20+ years rotating shift work | Nurses | Age, age at menarche, menopausal status, age at menopause, age at first birth and parity combined, age at first birth, BMI, current alcohol consumption, oral contraceptive use, postmenopausal hormone use, smoking status, benign breast disease, family history of breast cancer, and physical activity | Questionnaire |
| Pronk A | 2010 | 349/69,472 | 0.8 (0.5–1.2) | Cohort study | 8 | China | Breast cancer | Never vs. 17+ years | NA | Age, education, family history of breast cancer, number of pregnancies, age at first birth, and occupational physical activity | Interview |
| Schernhammer ES | 2003 | 602/78,586 | 1.35 (1.03–1.77) | Cohort study | 8 | USA | Colorectal cancer | Never vs. 15+ years rotating shift work | Nurses | Age in years; pack-years of smoking before age 30 in quintiles; BMI in five categories; physical activity in quintiles; regular aspirin use; colorectal cancer in parent or sibling; screening endoscopy during the study period; consumption of beef, pork, or lamb as a main dish; alcohol consumption status; total caloric intake in quintiles; use of postmenopausal hormones; menopausal status; and height in seven categories | Questionnaire |
| Vistisen HT | 2014 | 1,281/169,011 | 1.68 (1.28–2.2) | Cohort study | 5 | Denmark | Breast cancer | Never vs. ever | NA | Shift work | Database |
| Schernhammer ES | 2013 | 1,455/78,612 | 1.28 (1.07–1.53) | Cohort study | 7 | USA | Lung cancer | Never vs. 15+ years rotating shift work | Nurses | Age, age at start smoking, cigarettes smoked per day, fruit intake, vegetable intake, BMI, measured as weight in kilograms divided by height in meters squared, menopausal status, hormone use among postmenopausal women, and oral contraceptive use, as well as environmental smoking exposures: parents smoking while living with them, years living with someone who smoked, exposure to smoking at work, and exposure to smoking at home | Questionnaire |
| Gu F | 2015 | 5,413/74,862 | 1.08 (0.98–1.08) | Cohort study | 7 | USA | Unclassified cancer | Never vs. 15+ years rotating shift work | Nurses | Age, alcohol consumption, physical exercise, multivitamin use, menopausal status and postmenopausal hormone use, physical exam in the past 2 years, healthy eating score (quintiles), smoking status, pack-years; BMI, and husband's education | Questionnaire |
| Ren Z | 2014 | 712/1,454 | 1.34 (1.04–1.71) | Case–control | 6 | China | Breast cancer | Never vs. ever | NA | Age, education, BMI, marital status, age at menarche, menopausal status, parity, activity, breastfeeding, family history of breast cancer, and other sleep factors | Database |
| Santi SA | 2015 | 743/1,518 | 1.7 (1.04–2.79) | Case–control | 7 | Canada | Breast cancer | Never vs. 10+ years rotating shift work | Nurses | Age, family history, level of education, oral contraception use, alcohol consumption, number of births, and age of first menstruation | Questionnaire |
| Datta K | 2014 | 50/150 | 1.509 (0.267–8.516) | Case–control | 6 | South Asia | Breast cancer | Never vs. ever | NA | Working in night shift | Direct interview |
| Lie JS | 2013 | 172/646 | 1.8 (1–3.1) | Case–control | 7 | Norway | ER+ breast cancer | Never vs. 5+ years night shift work | Nurses | Age at diagnosis, period of diagnosis, parity, family history of breast cancer in mother or sister, hormonal treatment in the previous 2 years before diagnosis, and frequency of alcohol consumption at the time of diagnosis | Telephone interview |
| Lie JS | 2013 | 22/346 | 2.8 (0.8–9.2) | Case–control | 7 | Norway | ER− breast cancer | Never vs. 5+ years night shift work | Nurses | Age at diagnosis, period of diagnosis, parity, family history of breast cancer in mother or sister, hormonal treatment in the previous 2 years before diagnosis, and frequency of alcohol consumption at the time of diagnosis | Telephone interview |
| Lie JS | 2006 | 537/2,680 | 2.21 (1.1–4.45) | Nested case–control | 6 | Norway | Breast cancer | Never vs. 30+ years night shift work | Nurses | Total employment time as a nurses and parity | Database |
| Papantoniou K | 2015 | 1,708/3,486 | 1.22 (0.82–1.81) | Case–control | 8 | Spain | Breast cancer | Never vs. 15+ years night shift work | NA | Age, center, educational level, parity, menopausal status, family history of breast cancer, BMI, smoking status, oral contraceptive use, leisure time physical activity, alcohol consumption, sleep duration | Face-to-face interview |
| Truong T | 2014 | 1,126/2,300 | 1.32 (1.02–1.72) | Case–control | 8 | France | Breast cancer | Never vs. ever | NA | Age, study area, parity, age at first full-term pregnancy, age at menarche, family history of breast cancer, current use of hormonal replacement therapy, BMI, and tobacco and alcohol consumption | In-person interview |
| Hansen J | 2012 | 267/1,302 | 2.1 (1.3–3.2) | Nested case–control | 7 | Denmark | Breast cancer | Never vs. 20+ years night shift work | Nurses | Age, weight regularity, use of hormone replacement therapy, age at menarche, menstrual regularity, menopausal status, age at birth of first child, breast cancer in mother or sister, total duration of lactation | Interview |
| Kloog I | 2011 | 794/1,679 | 1.359 (1.121–1.647) | Case–control | 6 | Northern Israel | Breast cancer | No light vs. light at night | NA | Light at night | Interview |
| Hansen J | 2001 | 5,964/11,750 | 1.7 (1.3–1.7) | Case–control | 8 | Denmark | Breast cancer | Never vs. 6+ years night shift work | NA | Age, social class, age at birth of first child, age at birth of last child, and number of children | Interview |
| Grundy A | 2013 | 1,134/2,313 | 2.21 (1.14–4.31) | Case–control | 7 | Canada | Breast cancer | Never vs. 30+ years night shift work | NA | Years of night shift history | Questionnaire |
| Li Q | 2010 | 74/201 | 0.9 (0.2–3.9) | Case–control | 8 | USA | Breast cancer | Never vs. ever | NA | Age group, race, BMI, age at first menstrual period, family history of breast cancer, age at first full-term birth, months of lifetime breast feeding, cigarette smoking, alcohol drinking, and premenopausal | Interview |
| Li Q | 2010 | 289/518 | 1.4 (0.5–4.3) | Case–control | 8 | USA | Breast cancer | Never vs. ever | NA | Age group, race, BMI, age at first menstrual period, family history of breast cancer, age at first full-term birth, months of lifetime breast feeding, cigarette smoking, alcohol drinking, and postmenopausal | Interview |
| Hansen J | 2012 | 218/1,117 | 2.1 (1–4.5) | Nested case–control | 7 | Denmark | Breast cancer | Never vs. 15+ years night shift work | Military women | Age, hormone replacement therapy, number of childbirths, age at menarche, years of education, occasional sunbathing frequency, and tobacco smoking status | Questionnaire |
| Kwon P | 2015 | 1,451/4,491 | 0.88 (0.69–1.12) | Nested case–control | 7 | China | Breast cancer | Never vs. 30.6+ years night shift work | Textile workers | Age, smoking, parity, and endotoxin | Factory personnel record review (80%), supervisor interviews (12%), and in-person employee or close relative interviews (8%). |
| Davis S | 2001 | 813/1,606 | 2.3 (1.2–4.2) | Case–control | 8 | USA | Breast cancer | Never vs. 4.6+ years night shift work | NA | Parity, family history of breast cancer, oral contraceptive use, and recent discontinued use of hormone replacement therapy | In-person interview |
| Rabstein S | 2013 | 857/1,749 | 1.01 (0.68–1.5) | Case–control | 8 | Germany | Breast cancer | Never vs. ever | NA | Family history of breast cancer, hormone replacement use, number of mammograms, and estrogen receptor status | Telephone interview |
| Menegaux F | 2013 | 1,232/2,549 | 1.4 (1.01–1.92) | Case–control | 8 | France | Breast cancer | Never vs. 4.5+ years night shift work | NA | Age, study area, parity, age at first full-term pregnancy, age at menarche, family history of breast cancer, current hormonal replacement therapy, BMI, tobacco and alcohol | In-person interview |
| Bhatti P | 2013 | 1,101/2,933 | 1.02 (0.74–1.42) | Case–control | 8 | USA | Invasive epithelial ovarian cancer | Never vs. 7+ years night shift work | NA | Age at reference, county, reference year, duration of oral contraceptive use, number of full-term pregnancies, and BMI at age 30 | In-person interview |
| Wang P | 2015 | 712/1,454 | 1.34 (1.05–1.72) | Case–control | 8 | China | Breast cancer | Never vs. ever | NA | Age, education, BMI, age at menarche, menopausal status, parity, physical activity, breast-feeding, family history of breast cancer, and other sleep factors | Face-to-face interview |
| Li W | 2015 | 1,709/6,489 | 0.73 (0.66–0.82) | Nested case–control | 6 | China | Breast cancer | Never vs. ever | Textile workers | Age at the beginning of follow-up | Review of factory personnel records (80%), interviews of factory supervisors (12%), and in-person interviews of women or their relatives |
| O'Leary ES | 2006 | 486/1,444 | 0.32 (0.12–0.83) | Case–control | 8 | USA | Breast cancer | Never vs. 8+ years night shift work | NA | Age at reference date, parity, family history, education, and history of benign breast disease | 30-minute in-home interview |
| Fritschi L | 2013 | 1,202/2,987 | 1.02 (0.71–1.45) | Case–control | 7 | Australia | Breast cancer | Never vs. 20+ years night shift work | NA | Night shift work | Questionnaire and a follow-up telephone interview |
| Schwartzbaum J | 2007 | 236/1148,661 | 1 (0.89–1.13) | Cohort study | 8 | Sweden | bCancer | Ever vs. never | NA | Age, socioeconomic status, occupational position, and county of residence | Interview |
| Bauer SE | 2013 | 34,053/48,511 | 1.12 (1.04–1.2) | Case–control | 7 | USA | Breast cancer | Low light at night vs. high | NA | Race, tumor grade and stage, year of diagnosis, age at cancer diagnosis, Metropolitan Statistical Area (MSA) status, births per 1,000 women ages 15–50, MSA population mobility, population over 16 in the labor force, and prevalence of cigarette smoking | Infrared light detection by a nighttime satellite |
| Kojo K | 2005 | 27/544 | 1.52 (0.49–4.74) | Nested case–control | 7 | Finland | Breast cancer | Never vs. often | Cabin attendants | Cumulative radiation dose, number of fertile years, parity, family history of breast cancer, alcohol consumption, disruption of sleep rhythm, and menstrual cycle | Questionnaire |
| Rafnsson V | 2003 | 35/175 | 5.24 (1.58–17.38) | Nested case–control | 6 | Iceland | Breast cancer | <5 years vs. ≥ 5 years | Cabin attendants | Age at first childbirth and length of employment, number of children | Employment history |
| Rafnsson V | 2003 | 35/175 | 0.82 (0.34–1.97) | Nested case–control | 6 | Iceland | Breast cancer | <5 years vs. ≥ 5 years | Cabin attendants | Age at first childbirth and length of employment, number of children | Employment history |
| Pukkala E | 2012 | 559/8,244 | 1.16 (1.06–1.25) | Nested case–control | 5 | Finland, Iceland, Norway and Sweden | b Cancer | Cabin crew vs. general population | Airline cabin crew | Length of employment | Employment history |
| Linnersjö A | 2003 | 71/2,324 | 1.01 (0.78–1.24) | Nested case–control | 5 | Sweden | b Cancer | Cabin crew vs. general population | Airline cabin crew | Age and calendar period | Employment history |
| Reynolds P | 2002 | 104/44,021 | 1.05 (0.86–1.27) | Cohort study | 5 | USA | Unclassified cancer | Flight attendants versus general population | Flight attendants | None | Historical AFA flight records extrapolated to California cohort |
| McElroy JA | 2006 | 4,033/9,347 | 0.94 (0.62–1.44) | Case–control | 8 | USA | Breast cancer | Short duration of sleep per night vs. normal | NA | Reference age, state, parity, age at first full-term pregnancy, family history of breast cancer, alcohol consumption, BMI, menopausal status, age at menopause, postmenopausal hormone use, education, and marital status | Interviews |
| Schernhammer E | 2014 | 9,140/193,396 | 0.95 (0.77–1.17 | Cohort study | 5 | USA | Breast cancer | Never vs. 30+ years rotating shift work | Nurses | Multivariable adjustment | Large prospective data sets |
| Marino JL | 2008 | 812/2,125 | 1.2 (1–1.5) | Case–control | 5 | USA | Epithelial ovarian cancer | Never ns ever | NA | Multivariable adjustment | In-person interviews |
| Lie JS | 2011 | 699/1,594 | 1.3 (0.9–1.8) | Nested case–control study | 8 | Norway | Breast cancer | Never vs. 12+ years rotating shift work | Nurses | Age, period of diagnosis, parity, family history of breast cancer in mother or sister (no/yes), and frequency of alcohol consumption at time of diagnosis | Telephone interviews |
| Lie S | 2013 | 513/1,270 | 2.4 (1.3–4.3) | Case–control | 5 | Norway | Breast cancer | Never vs. 5+ years night shift work | Nurses | None | Database |
| Lahti TA | 2008 | 2,494/NAa | 1.02 (0.94–1.12) | Cohort study | 6 | Finland | Non-Hodgkin lymphoma | Never vs. ever | NA | Age, social class, and cohort period | Finnish Job-Exposure Matrix (FINJEM) |
| Girschik J | 2013 | 624/1,543 | 1.05 (0.82–1.33) | Case–control | 8 | Australia | Breast cancer | Short duration of sleep t vs. normal | NA | Age, number of children, age at first birth, breastfeeding, menopausal status, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, and physical activity | Questionnaire |
| Tsai RJ | 2014 | 839/2,457 | 1.34 (1.06–1.68) | Case–control | 8 | USA | Breast cancer | Daytime work vs. regular rotating shift work | NA | Obesity, smoking status, alcohol consumption, race, income, education, health insurance coverage, and marital status | Interview |
| Tsai RJ | 2014 | 1,253/6,238 | 0.98 (0.78–1.21) | Case–control | 8 | USA | Cervical cancer | Daytime work vs. regular rotating shift work | NA | Obesity, smoking status, alcohol consumption, race, income, education, health insurance coverage, and marital status | Interview |
| Tsai RJ | 2014 | 1,412/2,176 | 1.17 (1.01–1.36) | Case–control | 8 | USA | Colon cancer | Daytime work vs. regular rotating shift work | NA | Obesity, smoking status, alcohol consumption, race, income, education, health insurance coverage, and marital status | Interview |
| Chu CH | 2010 | 408/2,023 | 2.54 (1.37–4.7) | Nested case–control study | 5 | China | Breast cancer | Never vs. ever | NA | Potential confounders | Interview |
| Luo J | 2012 | 295/14,2933 | 0.79 (0.61–1.02) | Cohort study | 8 | USA | Thyroid cancer | Short sleep duration vs. normal | NA | Age at enrollment, ethnicity, educational level, smoking, BMI (weight (kg)/height (m)2), recreational physical activity, alcohol intake, family history of cancer, previous thyroid disease, history of hormone therapy use, depression score, and different treatment assignments for Women's Health Initiative clinical trials | Questionnaire |
| Verkasalo PK | 2005 | 242/12,222 | 0.88 (0.11–6.91) | Cohort study | 8 | Finland | Breast cancer | Short sleep duration vs. normal | NA | Age, zygosity, social class, number of children, use of oral contraceptives, BMI, alcohol use, smoking, and physical activity | Questionnaire |
| Jiao L | 2013 | 851/7,5828 | 1.36 (1.06–1.74) | Cohort study | 8 | USA | Colorectal cancer | Short sleep duration vs. normal | NA | Age, ethnicity, fatigue, hormone replacement therapy, waist-to-hip ratio, and physical activity | Questionnaire |
| Kakizaki M | 2008 | 143/23,995 | 1.67 (1.002–2.78) | Cohort study | 8 | Japan | Breast cancer | Short sleep duration vs. normal | NA | Age, BMI, history of diseases, family history of cancer, job, marital status, education, cigarette smoking, alcohol consumption, time spent walking, total caloric intake, menopausal status, age at menarche, age at first delivery, number of deliveries, use of oral contraceptive drugs, use of hormone drugs except for oral contraceptive drugs (yes or no) | Questionnaire |
| Tynes T | 1996 | 140/2,169 | 1.2 (1–1.4) | Cohort study | 6 | Norway | b Cancer | Never vs. ever | Radio and telegraph operators | Shift work and duration of employment | Database |
| Pesch B | 2010 | 749/1,542 | 2.48 (0.62–9.99) | Case–control | 8 | Germany | Breast cancer | Never vs. 20+ years night shift work | NA | A potential selection bias using bootstrapping, family history of breast cancer, hormone replacement use, and number of mammograms | Interview |
| Bai Y | 2016 | 613/14004 | 0.90 (0.66–1.23) | Cohort study | 8 | China | Unclassified cancer | Never vs. 20+ years night shift work | NA | Age, BMI, family history of cancer, alcohol drinking and smoking status, number of children, menopausal status, hormone replacement therapy, and contraception status | Questionnaire |
| Cohen JM | 2015 | 415/10,2484 | 0.90 (0.67–1.20) | Cohort study | 8 | USA | Melanoma | Short sleep duration vs. normal | Nurses | Age, number of sunburns, moles, hair color, family history of melanoma (yes, no), reaction to the sun, tanning, Caucasian ethnicity, ultraviolet flux (quintiles), snoring | Questionnaire |
| Travis RC | 2016 | 4,809/522,246 | 1.00 (0.92–1.08) | Cohort study | 8 | UK | Breast cancer | Never vs. ever | NA | Socioeconomic status, parity and age at first birth, BMI, alcohol intake, strenuous physical activity, family history of breast cancer, age at menarche, oral contraceptive use, smoking, living, with a partner, and use of menopausal hormone therapy | Questionnaire |
| Wegrzyn LR (NHS)c | 2017 | 5,971/7,8516 | 0.95 (0.77–1.17) | Cohort study | 8 | USA | Breast cancer | Never vs. 30+ years night shift work | Nurses | Age, height, BMI, BMI at age 18, adolescent body size, age at menarche, age at first birth and parity combined, breastfeeding, type of menopause and age at menopause combined, menopausal hormone therapy, duration of estrogen alone menopausal hormone therapy, duration of estrogen and progesterone menopausal hormone therapy, first-degree family history of breast cancer, history of benign breast disease, alcohol consumption, physical activity, and current mammography use | Questionnaire |
| Wegrzyn LR (NHS2)c | 2017 | 3,570/114,559 | 2.15 (1.23–3.73) | Cohort study | 8 | USA | Breast cancer | Never vs. 20+ years night shift work | Nurses | Age, height, BMI, BMI at age 18, adolescent body size, age at menarche, age at first birth and parity combined, breastfeeding, type of menopause and age at menopause combined, menopausal hormone therapy, duration of estrogen alone menopausal hormone therapy, duration of estrogen and progesterone menopausal hormone therapy, first-degree family history of breast cancer, history of benign breast disease, alcohol consumption, physical activity, and current mammography use | Questionnaire |
| Study | Published year | No. of cases/No. of subjects | OR (95% CI) | Study design | Quality score | Region | Type of cancer | Range of night shift work | Occupation | Variables of adjustment | Exposure assessment method |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Knutsson A | 2012 | 94/4,036 | 2.02 (1.03–3.95) | Cohort study | 7 | Sweden | Breast cancer | Day vs. night shift | NA | Number of children, alcohol consumption, BMI, height, weight, waist, hip circumference, educational level, smoking menopausal status, status of oral contraceptive use, and hormones other than contraceptives | Questionnaire |
| Carter BD | 2014 | 1,289/16,1004 | 1.27 (1.03–1.56) | Cohort study | 8 | USA | Ovarian cancer | Day vs. rotating shifts | NA | Oral contraceptive use, age at menarche and menopause, tubal ligation, parity, postmenopausal estrogen use, race, family history of breast/ovarian cancers, exercise, BMI, and height | Questionnaire |
| Poole EM | 2011 | 718/181,548 | 0.8 (0.51–1.23) | Cohort study | 7 | USA | Ovarian cancer | Never vs. 20+ years rotating shift work | Nurses | Age, duration of oral contraceptive use, parity, BMI, smoking status, tubal ligation history, menopausal status, family history of ovarian cancer, duration of breas tfeeding, and cohort | Questionnaire |
| Viswanathan AN | 2007 | 515/53,487 | 1.47 (1.03–2.1) | Cohort study | 7 | USA | Endometrial cancer | Never vs. 20+ years rotating shift work | Nurses | Age, age at menarche, age at menopause, parity, BMI duration of oral contraceptive use, use and duration of postmenopausal hormones, hypertension, diabetes, pack-years of smoking | Questionnaire |
| Akerstedt T | 2015 | 463/1,3656 | 1.77 (1.03–3.04) | Cohort study | 8 | Sweden | Breast cancer | Never vs. 21–45 years | NA | Age, education level, tobacco consumption, BMI, having children, coffee consumption, previous cancer, use of hormones including oral contraceptives | Telephone interview |
| Koppes LLJ | 2014 | 2,531/285,723 | 0.87 (0.72–1.05) | Cohort study | 8 | Netherland | Breast cancer | Never vs. regular night work | NA | Age, origin, children in household education, occupation, job tenure (years) | Personal interview |
| Natti J | 2012 | 48/1,649 | 2.82 (1.196–6.645) | Cohort study | 8 | Finland | Unclassified cancer | Day vs. weekly night shift | NA | Age, and smoking status, demographics and health- and work-related factors | Face-to-face interview |
| Schernhammer ES | 2006 | 1,352/115,022 | 1.79 (1.06–3.01) | Cohort study | 7 | USA | Breast cancer | Never vs. 20+ years rotating shift work | Nurses | Age, age at menarche, menopausal status, age at menopause, age at first birth and parity combined, age at first birth, BMI, current alcohol consumption, oral contraceptive use, postmenopausal hormone use, smoking status, benign breast disease, family history of breast cancer, and physical activity | Questionnaire |
| Pronk A | 2010 | 349/69,472 | 0.8 (0.5–1.2) | Cohort study | 8 | China | Breast cancer | Never vs. 17+ years | NA | Age, education, family history of breast cancer, number of pregnancies, age at first birth, and occupational physical activity | Interview |
| Schernhammer ES | 2003 | 602/78,586 | 1.35 (1.03–1.77) | Cohort study | 8 | USA | Colorectal cancer | Never vs. 15+ years rotating shift work | Nurses | Age in years; pack-years of smoking before age 30 in quintiles; BMI in five categories; physical activity in quintiles; regular aspirin use; colorectal cancer in parent or sibling; screening endoscopy during the study period; consumption of beef, pork, or lamb as a main dish; alcohol consumption status; total caloric intake in quintiles; use of postmenopausal hormones; menopausal status; and height in seven categories | Questionnaire |
| Vistisen HT | 2014 | 1,281/169,011 | 1.68 (1.28–2.2) | Cohort study | 5 | Denmark | Breast cancer | Never vs. ever | NA | Shift work | Database |
| Schernhammer ES | 2013 | 1,455/78,612 | 1.28 (1.07–1.53) | Cohort study | 7 | USA | Lung cancer | Never vs. 15+ years rotating shift work | Nurses | Age, age at start smoking, cigarettes smoked per day, fruit intake, vegetable intake, BMI, measured as weight in kilograms divided by height in meters squared, menopausal status, hormone use among postmenopausal women, and oral contraceptive use, as well as environmental smoking exposures: parents smoking while living with them, years living with someone who smoked, exposure to smoking at work, and exposure to smoking at home | Questionnaire |
| Gu F | 2015 | 5,413/74,862 | 1.08 (0.98–1.08) | Cohort study | 7 | USA | Unclassified cancer | Never vs. 15+ years rotating shift work | Nurses | Age, alcohol consumption, physical exercise, multivitamin use, menopausal status and postmenopausal hormone use, physical exam in the past 2 years, healthy eating score (quintiles), smoking status, pack-years; BMI, and husband's education | Questionnaire |
| Ren Z | 2014 | 712/1,454 | 1.34 (1.04–1.71) | Case–control | 6 | China | Breast cancer | Never vs. ever | NA | Age, education, BMI, marital status, age at menarche, menopausal status, parity, activity, breastfeeding, family history of breast cancer, and other sleep factors | Database |
| Santi SA | 2015 | 743/1,518 | 1.7 (1.04–2.79) | Case–control | 7 | Canada | Breast cancer | Never vs. 10+ years rotating shift work | Nurses | Age, family history, level of education, oral contraception use, alcohol consumption, number of births, and age of first menstruation | Questionnaire |
| Datta K | 2014 | 50/150 | 1.509 (0.267–8.516) | Case–control | 6 | South Asia | Breast cancer | Never vs. ever | NA | Working in night shift | Direct interview |
| Lie JS | 2013 | 172/646 | 1.8 (1–3.1) | Case–control | 7 | Norway | ER+ breast cancer | Never vs. 5+ years night shift work | Nurses | Age at diagnosis, period of diagnosis, parity, family history of breast cancer in mother or sister, hormonal treatment in the previous 2 years before diagnosis, and frequency of alcohol consumption at the time of diagnosis | Telephone interview |
| Lie JS | 2013 | 22/346 | 2.8 (0.8–9.2) | Case–control | 7 | Norway | ER− breast cancer | Never vs. 5+ years night shift work | Nurses | Age at diagnosis, period of diagnosis, parity, family history of breast cancer in mother or sister, hormonal treatment in the previous 2 years before diagnosis, and frequency of alcohol consumption at the time of diagnosis | Telephone interview |
| Lie JS | 2006 | 537/2,680 | 2.21 (1.1–4.45) | Nested case–control | 6 | Norway | Breast cancer | Never vs. 30+ years night shift work | Nurses | Total employment time as a nurses and parity | Database |
| Papantoniou K | 2015 | 1,708/3,486 | 1.22 (0.82–1.81) | Case–control | 8 | Spain | Breast cancer | Never vs. 15+ years night shift work | NA | Age, center, educational level, parity, menopausal status, family history of breast cancer, BMI, smoking status, oral contraceptive use, leisure time physical activity, alcohol consumption, sleep duration | Face-to-face interview |
| Truong T | 2014 | 1,126/2,300 | 1.32 (1.02–1.72) | Case–control | 8 | France | Breast cancer | Never vs. ever | NA | Age, study area, parity, age at first full-term pregnancy, age at menarche, family history of breast cancer, current use of hormonal replacement therapy, BMI, and tobacco and alcohol consumption | In-person interview |
| Hansen J | 2012 | 267/1,302 | 2.1 (1.3–3.2) | Nested case–control | 7 | Denmark | Breast cancer | Never vs. 20+ years night shift work | Nurses | Age, weight regularity, use of hormone replacement therapy, age at menarche, menstrual regularity, menopausal status, age at birth of first child, breast cancer in mother or sister, total duration of lactation | Interview |
| Kloog I | 2011 | 794/1,679 | 1.359 (1.121–1.647) | Case–control | 6 | Northern Israel | Breast cancer | No light vs. light at night | NA | Light at night | Interview |
| Hansen J | 2001 | 5,964/11,750 | 1.7 (1.3–1.7) | Case–control | 8 | Denmark | Breast cancer | Never vs. 6+ years night shift work | NA | Age, social class, age at birth of first child, age at birth of last child, and number of children | Interview |
| Grundy A | 2013 | 1,134/2,313 | 2.21 (1.14–4.31) | Case–control | 7 | Canada | Breast cancer | Never vs. 30+ years night shift work | NA | Years of night shift history | Questionnaire |
| Li Q | 2010 | 74/201 | 0.9 (0.2–3.9) | Case–control | 8 | USA | Breast cancer | Never vs. ever | NA | Age group, race, BMI, age at first menstrual period, family history of breast cancer, age at first full-term birth, months of lifetime breast feeding, cigarette smoking, alcohol drinking, and premenopausal | Interview |
| Li Q | 2010 | 289/518 | 1.4 (0.5–4.3) | Case–control | 8 | USA | Breast cancer | Never vs. ever | NA | Age group, race, BMI, age at first menstrual period, family history of breast cancer, age at first full-term birth, months of lifetime breast feeding, cigarette smoking, alcohol drinking, and postmenopausal | Interview |
| Hansen J | 2012 | 218/1,117 | 2.1 (1–4.5) | Nested case–control | 7 | Denmark | Breast cancer | Never vs. 15+ years night shift work | Military women | Age, hormone replacement therapy, number of childbirths, age at menarche, years of education, occasional sunbathing frequency, and tobacco smoking status | Questionnaire |
| Kwon P | 2015 | 1,451/4,491 | 0.88 (0.69–1.12) | Nested case–control | 7 | China | Breast cancer | Never vs. 30.6+ years night shift work | Textile workers | Age, smoking, parity, and endotoxin | Factory personnel record review (80%), supervisor interviews (12%), and in-person employee or close relative interviews (8%). |
| Davis S | 2001 | 813/1,606 | 2.3 (1.2–4.2) | Case–control | 8 | USA | Breast cancer | Never vs. 4.6+ years night shift work | NA | Parity, family history of breast cancer, oral contraceptive use, and recent discontinued use of hormone replacement therapy | In-person interview |
| Rabstein S | 2013 | 857/1,749 | 1.01 (0.68–1.5) | Case–control | 8 | Germany | Breast cancer | Never vs. ever | NA | Family history of breast cancer, hormone replacement use, number of mammograms, and estrogen receptor status | Telephone interview |
| Menegaux F | 2013 | 1,232/2,549 | 1.4 (1.01–1.92) | Case–control | 8 | France | Breast cancer | Never vs. 4.5+ years night shift work | NA | Age, study area, parity, age at first full-term pregnancy, age at menarche, family history of breast cancer, current hormonal replacement therapy, BMI, tobacco and alcohol | In-person interview |
| Bhatti P | 2013 | 1,101/2,933 | 1.02 (0.74–1.42) | Case–control | 8 | USA | Invasive epithelial ovarian cancer | Never vs. 7+ years night shift work | NA | Age at reference, county, reference year, duration of oral contraceptive use, number of full-term pregnancies, and BMI at age 30 | In-person interview |
| Wang P | 2015 | 712/1,454 | 1.34 (1.05–1.72) | Case–control | 8 | China | Breast cancer | Never vs. ever | NA | Age, education, BMI, age at menarche, menopausal status, parity, physical activity, breast-feeding, family history of breast cancer, and other sleep factors | Face-to-face interview |
| Li W | 2015 | 1,709/6,489 | 0.73 (0.66–0.82) | Nested case–control | 6 | China | Breast cancer | Never vs. ever | Textile workers | Age at the beginning of follow-up | Review of factory personnel records (80%), interviews of factory supervisors (12%), and in-person interviews of women or their relatives |
| O'Leary ES | 2006 | 486/1,444 | 0.32 (0.12–0.83) | Case–control | 8 | USA | Breast cancer | Never vs. 8+ years night shift work | NA | Age at reference date, parity, family history, education, and history of benign breast disease | 30-minute in-home interview |
| Fritschi L | 2013 | 1,202/2,987 | 1.02 (0.71–1.45) | Case–control | 7 | Australia | Breast cancer | Never vs. 20+ years night shift work | NA | Night shift work | Questionnaire and a follow-up telephone interview |
| Schwartzbaum J | 2007 | 236/1148,661 | 1 (0.89–1.13) | Cohort study | 8 | Sweden | bCancer | Ever vs. never | NA | Age, socioeconomic status, occupational position, and county of residence | Interview |
| Bauer SE | 2013 | 34,053/48,511 | 1.12 (1.04–1.2) | Case–control | 7 | USA | Breast cancer | Low light at night vs. high | NA | Race, tumor grade and stage, year of diagnosis, age at cancer diagnosis, Metropolitan Statistical Area (MSA) status, births per 1,000 women ages 15–50, MSA population mobility, population over 16 in the labor force, and prevalence of cigarette smoking | Infrared light detection by a nighttime satellite |
| Kojo K | 2005 | 27/544 | 1.52 (0.49–4.74) | Nested case–control | 7 | Finland | Breast cancer | Never vs. often | Cabin attendants | Cumulative radiation dose, number of fertile years, parity, family history of breast cancer, alcohol consumption, disruption of sleep rhythm, and menstrual cycle | Questionnaire |
| Rafnsson V | 2003 | 35/175 | 5.24 (1.58–17.38) | Nested case–control | 6 | Iceland | Breast cancer | <5 years vs. ≥ 5 years | Cabin attendants | Age at first childbirth and length of employment, number of children | Employment history |
| Rafnsson V | 2003 | 35/175 | 0.82 (0.34–1.97) | Nested case–control | 6 | Iceland | Breast cancer | <5 years vs. ≥ 5 years | Cabin attendants | Age at first childbirth and length of employment, number of children | Employment history |
| Pukkala E | 2012 | 559/8,244 | 1.16 (1.06–1.25) | Nested case–control | 5 | Finland, Iceland, Norway and Sweden | b Cancer | Cabin crew vs. general population | Airline cabin crew | Length of employment | Employment history |
| Linnersjö A | 2003 | 71/2,324 | 1.01 (0.78–1.24) | Nested case–control | 5 | Sweden | b Cancer | Cabin crew vs. general population | Airline cabin crew | Age and calendar period | Employment history |
| Reynolds P | 2002 | 104/44,021 | 1.05 (0.86–1.27) | Cohort study | 5 | USA | Unclassified cancer | Flight attendants versus general population | Flight attendants | None | Historical AFA flight records extrapolated to California cohort |
| McElroy JA | 2006 | 4,033/9,347 | 0.94 (0.62–1.44) | Case–control | 8 | USA | Breast cancer | Short duration of sleep per night vs. normal | NA | Reference age, state, parity, age at first full-term pregnancy, family history of breast cancer, alcohol consumption, BMI, menopausal status, age at menopause, postmenopausal hormone use, education, and marital status | Interviews |
| Schernhammer E | 2014 | 9,140/193,396 | 0.95 (0.77–1.17 | Cohort study | 5 | USA | Breast cancer | Never vs. 30+ years rotating shift work | Nurses | Multivariable adjustment | Large prospective data sets |
| Marino JL | 2008 | 812/2,125 | 1.2 (1–1.5) | Case–control | 5 | USA | Epithelial ovarian cancer | Never ns ever | NA | Multivariable adjustment | In-person interviews |
| Lie JS | 2011 | 699/1,594 | 1.3 (0.9–1.8) | Nested case–control study | 8 | Norway | Breast cancer | Never vs. 12+ years rotating shift work | Nurses | Age, period of diagnosis, parity, family history of breast cancer in mother or sister (no/yes), and frequency of alcohol consumption at time of diagnosis | Telephone interviews |
| Lie S | 2013 | 513/1,270 | 2.4 (1.3–4.3) | Case–control | 5 | Norway | Breast cancer | Never vs. 5+ years night shift work | Nurses | None | Database |
| Lahti TA | 2008 | 2,494/NAa | 1.02 (0.94–1.12) | Cohort study | 6 | Finland | Non-Hodgkin lymphoma | Never vs. ever | NA | Age, social class, and cohort period | Finnish Job-Exposure Matrix (FINJEM) |
| Girschik J | 2013 | 624/1,543 | 1.05 (0.82–1.33) | Case–control | 8 | Australia | Breast cancer | Short duration of sleep t vs. normal | NA | Age, number of children, age at first birth, breastfeeding, menopausal status, use of hormone replacement therapy, duration of use of hormone replacement therapy, alcohol consumption, comparative weight at age 30 years, ever use of melatonin, and physical activity | Questionnaire |
| Tsai RJ | 2014 | 839/2,457 | 1.34 (1.06–1.68) | Case–control | 8 | USA | Breast cancer | Daytime work vs. regular rotating shift work | NA | Obesity, smoking status, alcohol consumption, race, income, education, health insurance coverage, and marital status | Interview |
| Tsai RJ | 2014 | 1,253/6,238 | 0.98 (0.78–1.21) | Case–control | 8 | USA | Cervical cancer | Daytime work vs. regular rotating shift work | NA | Obesity, smoking status, alcohol consumption, race, income, education, health insurance coverage, and marital status | Interview |
| Tsai RJ | 2014 | 1,412/2,176 | 1.17 (1.01–1.36) | Case–control | 8 | USA | Colon cancer | Daytime work vs. regular rotating shift work | NA | Obesity, smoking status, alcohol consumption, race, income, education, health insurance coverage, and marital status | Interview |
| Chu CH | 2010 | 408/2,023 | 2.54 (1.37–4.7) | Nested case–control study | 5 | China | Breast cancer | Never vs. ever | NA | Potential confounders | Interview |
| Luo J | 2012 | 295/14,2933 | 0.79 (0.61–1.02) | Cohort study | 8 | USA | Thyroid cancer | Short sleep duration vs. normal | NA | Age at enrollment, ethnicity, educational level, smoking, BMI (weight (kg)/height (m)2), recreational physical activity, alcohol intake, family history of cancer, previous thyroid disease, history of hormone therapy use, depression score, and different treatment assignments for Women's Health Initiative clinical trials | Questionnaire |
| Verkasalo PK | 2005 | 242/12,222 | 0.88 (0.11–6.91) | Cohort study | 8 | Finland | Breast cancer | Short sleep duration vs. normal | NA | Age, zygosity, social class, number of children, use of oral contraceptives, BMI, alcohol use, smoking, and physical activity | Questionnaire |
| Jiao L | 2013 | 851/7,5828 | 1.36 (1.06–1.74) | Cohort study | 8 | USA | Colorectal cancer | Short sleep duration vs. normal | NA | Age, ethnicity, fatigue, hormone replacement therapy, waist-to-hip ratio, and physical activity | Questionnaire |
| Kakizaki M | 2008 | 143/23,995 | 1.67 (1.002–2.78) | Cohort study | 8 | Japan | Breast cancer | Short sleep duration vs. normal | NA | Age, BMI, history of diseases, family history of cancer, job, marital status, education, cigarette smoking, alcohol consumption, time spent walking, total caloric intake, menopausal status, age at menarche, age at first delivery, number of deliveries, use of oral contraceptive drugs, use of hormone drugs except for oral contraceptive drugs (yes or no) | Questionnaire |
| Tynes T | 1996 | 140/2,169 | 1.2 (1–1.4) | Cohort study | 6 | Norway | b Cancer | Never vs. ever | Radio and telegraph operators | Shift work and duration of employment | Database |
| Pesch B | 2010 | 749/1,542 | 2.48 (0.62–9.99) | Case–control | 8 | Germany | Breast cancer | Never vs. 20+ years night shift work | NA | A potential selection bias using bootstrapping, family history of breast cancer, hormone replacement use, and number of mammograms | Interview |
| Bai Y | 2016 | 613/14004 | 0.90 (0.66–1.23) | Cohort study | 8 | China | Unclassified cancer | Never vs. 20+ years night shift work | NA | Age, BMI, family history of cancer, alcohol drinking and smoking status, number of children, menopausal status, hormone replacement therapy, and contraception status | Questionnaire |
| Cohen JM | 2015 | 415/10,2484 | 0.90 (0.67–1.20) | Cohort study | 8 | USA | Melanoma | Short sleep duration vs. normal | Nurses | Age, number of sunburns, moles, hair color, family history of melanoma (yes, no), reaction to the sun, tanning, Caucasian ethnicity, ultraviolet flux (quintiles), snoring | Questionnaire |
| Travis RC | 2016 | 4,809/522,246 | 1.00 (0.92–1.08) | Cohort study | 8 | UK | Breast cancer | Never vs. ever | NA | Socioeconomic status, parity and age at first birth, BMI, alcohol intake, strenuous physical activity, family history of breast cancer, age at menarche, oral contraceptive use, smoking, living, with a partner, and use of menopausal hormone therapy | Questionnaire |
| Wegrzyn LR (NHS)c | 2017 | 5,971/7,8516 | 0.95 (0.77–1.17) | Cohort study | 8 | USA | Breast cancer | Never vs. 30+ years night shift work | Nurses | Age, height, BMI, BMI at age 18, adolescent body size, age at menarche, age at first birth and parity combined, breastfeeding, type of menopause and age at menopause combined, menopausal hormone therapy, duration of estrogen alone menopausal hormone therapy, duration of estrogen and progesterone menopausal hormone therapy, first-degree family history of breast cancer, history of benign breast disease, alcohol consumption, physical activity, and current mammography use | Questionnaire |
| Wegrzyn LR (NHS2)c | 2017 | 3,570/114,559 | 2.15 (1.23–3.73) | Cohort study | 8 | USA | Breast cancer | Never vs. 20+ years night shift work | Nurses | Age, height, BMI, BMI at age 18, adolescent body size, age at menarche, age at first birth and parity combined, breastfeeding, type of menopause and age at menopause combined, menopausal hormone therapy, duration of estrogen alone menopausal hormone therapy, duration of estrogen and progesterone menopausal hormone therapy, first-degree family history of breast cancer, history of benign breast disease, alcohol consumption, physical activity, and current mammography use | Questionnaire |
Abbreviations: NA, not available; BMI, body mass index.
aFrom the enrolled article, we could not obtain the accurate number of participants. Thus, in the preceding calculation, we missed the data. OR and corresponding 95% CI were extracted directly from the article.
bThese four studies reported the SIR for cancer overall among female night shifters and also reported on a number of different cancers, including, for example, breast cancer, colon cancer, cervix/uterus cancer, and other cancers.
cThis study examined associations between rotating night shift work and breast cancer in two prospective cohorts (NHS and NHS2).
Risk assessment and heterogeneity
We integrated multivariable-adjusted ORs of longest versus shortest exposure duration to identify the correlation between night shift work and the risk of common cancers. Long-term night shift work increased the risk of female cancer (OR = 1.190; 1.122–1.262) and OR forest plots (with 95% CIs) of cancer for long-term night shift female workers were shown in Supplementary Fig. S1. A positive association was observed regarding breast cancer (OR = 1.316; 95% CI, 1.196–1.448), digestive system cancer (OR = 1.177; 95% CI, 1.065–1.301) and skin cancer (OR = 1.408; 95% CI, 1.024–1.934). OR forest plots (with 95% CIs) of breast cancer, reproductive system cancers, digestive system cancers, lung cancer, and skin cancer were shown in Fig. 2. ORs of association between night shift work and the risk of common cancers in women were summarized in Supplementary Table S1. With regard to nurses, the risk of six common cancers were estimated and as a result night shift work was associated with increased risk of breast cancer (OR = 1.577; 95% CI, 1.235–2.014), digestive system cancer (OR = 1.350; 95% CI, 1.030–1.770), and lung cancer (OR = 1.280; 95% CI, 1.070–1.531). ORs of the common cancers in nurses were shown in Supplementary Table S2.
Forest plots of the association between night shift work and the risks of common cancers. I2, measure to quantify the degree of heterogeneity in meta-analyses. The squares and horizontal lines correspond to the study-specific OR and 95% CI. The area of the squares reflects the study-specific weight. The diamond represents the pooled OR and 95% CI.
Forest plots of the association between night shift work and the risks of common cancers. I2, measure to quantify the degree of heterogeneity in meta-analyses. The squares and horizontal lines correspond to the study-specific OR and 95% CI. The area of the squares reflects the study-specific weight. The diamond represents the pooled OR and 95% CI.
Significant heterogeneity was observed in the groups of breast cancer (P = 0.000, I2 = 80.4%), skin cancer (P = 0.009, I2 = 64.7%), and uterine cancer (P = 0.042, I2 = 59.6%), and risk estimates were conducted with a random-effect model. No evidence of heterogeneity existed among the other groups.
Subgroup analysis and meta-regression analysis
Subgroup analyses about breast cancer were conducted stratified by geographic location, study design, number of variables, study quality, exposure assessment, and occupation. When stratified by geographical location, Europe (OR = 1.416; 95% CI, 1.246–1.610) and North America (OR = 1.236; 95% CI, 1.048–1.459) shared higher OR estimates, whereas positive correlation did not exist in Asia (OR = 1.236; 95% CI, 0.865–1.767) and Australia (OR = 1.040; 95% CI, 0.852–1.271). When stratified by design, the highest pooled OR estimate was obtained among nested case–control studies (OR = 1.555; 95% CI, 1.115–2.169) and the lowest was among cohort studies (OR = 1.193; 95% CI, 1.030–1.382). In terms of occupation, it was revealed that the strongest OR existed in the nursing group (OR = 1.577; 1.235–2.014). In the analysis stratified by quality score, both groups of high score (OR = 1.273; 95% CI, 1.152–1.407) and low score (OR = 1.434; 95% CI, 1.133–1.814) presented a positive association but the latter tended to have significant heterogeneity (P = 0.000, I2 = 89.7%). As for the number of adjusted confounders, both groups indicated that night shift work was associated with increased risk of breast cancer and significant heterogeneity (P = 0.000, I2 = 89.1%) existed in the group adjusted for ≤3 confounders. When stratified by exposure assessment method, the questionnaire group presented a stronger correlation and lower heterogeneity compared with interview and database group, whereas negative association was observed among studies adopting other assessment methods (OR = 0.906; 95% CI, 0.596–1.378).
Furthermore, meta-regression analysis was carried out to explore the possible heterogeneity sources among stratified factors and none of the variables was considered as a potential source of heterogeneity. The results of subgroup analyses and meta-regression were summarized in Table 2.
Results of subgroup analyses and meta-regression analyses on the correlation between night shift work and the risk of breast cancer in women
| Risk estimates | Heterogeneity | ||||||
|---|---|---|---|---|---|---|---|
| Subgroup | Number of studiesa | Weight (%) | OR (95% CI) | P | P | I2 | Pinteraction |
| Geographic location | 0.472 | ||||||
| Europe | 27 (22, 26–27, 32, 38–44, 47, 50–51, 57, 59–62, 67–68, 74, 77–78, 81) | 54.42 | 1.416 (1.246–1.610) | 0.000 | 0.000 | 77.4% | |
| North America | 13 (29, 36, 45–46, 49, 55, 58, 64–65, 71, 82) | 25 | 1.236 (1.048–1.459) | 0.012 | 0.001 | 65.3% | |
| Asia | 7 (30, 35, 37, 53–54, 72, 76) | 15.13 | 1.236 (0.865–1.767) | 0.245 | 0.000 | 88.3% | |
| Australia | 2 (56, 70) | 5.45 | 1.040 (0.852–1.271) | 0.698 | 0.895 | 0.0% | |
| Study design | 0.584 | ||||||
| Cohort study | 14 (22, 26, 27, 29, 30, 32, 57, 65, 74, 76–77, 81, 82) | 32.32 | 1.193 (1.030–1.382) | 0.019 | 0.000 | 72.5% | |
| Case–control | 24 (35–38, 40–41, 43–46, 49–51, 53, 55–56, 58, 64, 68, 70–71, 78) | 47.79 | 1.329 (1.189–1.486) | 0.000 | 0.000 | 63.1% | |
| Nested case–control | 11 (39, 42, 47, 54, 59–62, 67, 72) | 19.89 | 1.555 (1.115–2.169) | 0.009 | 0.000 | 90.0% | |
| Number of variables | 0.592 | ||||||
| ≤3 | 15 (32, 37, 39, 43, 45, 54, 56, 60–62, 65, 68, 72, 77) | 31.18 | 1.440 (1.139–1.820) | 0.002 | 0.000 | 89.1% | |
| >3 | 34 (22, 26–27, 29–30, 35–36, 38, 40–42, 44, 46–47, 49–51, 53, 55, 57–59, 64, 67, 70–71, 74, 76, 78, 81, 82) | 68.82 | 1.273 (1.152–1.405) | 0.000 | 0.000 | 71.4% | |
| Study score | 0.658 | ||||||
| Low | 14 (32, 35, 37, 39, 43, 54, 60–62, 65, 68, 72, 77) | 30.37 | 1.434 (1.133–1.814) | 0.003 | 0.000 | 89.7% | |
| High | 35 (22, 26–27, 29–30, 36, 38, 40–42, 44–47, 49–51, 53, 55–59, 64, 67, 70–71, 74, 76, 78, 81, 82) | 69.63 | 1.273 (1.152–1.407) | 0.000 | 0.000 | 71.3% | |
| Exposure assessment | 0.077 | ||||||
| Database | 10 (32, 35, 39, 60–62, 65, 68, 77) | 21.85 | 1.452 (1.211–1.741) | 0.000 | 0.001 | 67.2% | |
| Questionnaire | 12 (22, 29, 36, 45, 47, 59, 70, 74, 76, 81, 82) | 21.42 | 1.380 (1.137–1.676) | 0.001 | 0.001 | 65.2% | |
| Interview | 25 (26–27, 30, 37–38, 40–44, 46, 49–51, 53, 55–57, 64, 67, 71–72, 78) | 49.54 | 1.287 (1.126–1.471) | 0.000 | 0.000 | 68.6% | |
| Other | 2 (54, 58) | 7.19 | 0.906 (0.596–1.378) | 0.645 | 0.000 | 97.6% | |
| Occupation | 0.350 | ||||||
| Nurses | 11 (29, 36, 38, 39, 42, 65, 67, 68, 82) | 21.02 | 1.577 (1.235–2.014) | 0.000 | 0.000 | 72.0% | |
| Unclassified occupation | 30 (22, 26–27, 30, 32, 35, 37, 40–41, 43–46, 49–51, 53, 55–58, 64, 70–72, 74, 76, 78, 81) | 63.60 | 1.250 (1.130–1.383) | 0.000 | 0.000 | 73.3% | |
| Flight attendants | 5 (59–62) | 7.98 | 1.454 (1.100–1.922) | 0.009 | 0.161 | 39.0% | |
| Military women | 1 (47) | 1.13 | 2.100 (0.990–4.455) | 0.053 | — | — | |
| Textile workers | 1 (54) | 3.55 | 0.730 (0.655–0.814) | 0.000 | — | — | |
| Radio and telegraph operators | 1 (77) | 2.73 | 1.500 (1.112–2.023) | 0.008 | — | — | |
| Risk estimates | Heterogeneity | ||||||
|---|---|---|---|---|---|---|---|
| Subgroup | Number of studiesa | Weight (%) | OR (95% CI) | P | P | I2 | Pinteraction |
| Geographic location | 0.472 | ||||||
| Europe | 27 (22, 26–27, 32, 38–44, 47, 50–51, 57, 59–62, 67–68, 74, 77–78, 81) | 54.42 | 1.416 (1.246–1.610) | 0.000 | 0.000 | 77.4% | |
| North America | 13 (29, 36, 45–46, 49, 55, 58, 64–65, 71, 82) | 25 | 1.236 (1.048–1.459) | 0.012 | 0.001 | 65.3% | |
| Asia | 7 (30, 35, 37, 53–54, 72, 76) | 15.13 | 1.236 (0.865–1.767) | 0.245 | 0.000 | 88.3% | |
| Australia | 2 (56, 70) | 5.45 | 1.040 (0.852–1.271) | 0.698 | 0.895 | 0.0% | |
| Study design | 0.584 | ||||||
| Cohort study | 14 (22, 26, 27, 29, 30, 32, 57, 65, 74, 76–77, 81, 82) | 32.32 | 1.193 (1.030–1.382) | 0.019 | 0.000 | 72.5% | |
| Case–control | 24 (35–38, 40–41, 43–46, 49–51, 53, 55–56, 58, 64, 68, 70–71, 78) | 47.79 | 1.329 (1.189–1.486) | 0.000 | 0.000 | 63.1% | |
| Nested case–control | 11 (39, 42, 47, 54, 59–62, 67, 72) | 19.89 | 1.555 (1.115–2.169) | 0.009 | 0.000 | 90.0% | |
| Number of variables | 0.592 | ||||||
| ≤3 | 15 (32, 37, 39, 43, 45, 54, 56, 60–62, 65, 68, 72, 77) | 31.18 | 1.440 (1.139–1.820) | 0.002 | 0.000 | 89.1% | |
| >3 | 34 (22, 26–27, 29–30, 35–36, 38, 40–42, 44, 46–47, 49–51, 53, 55, 57–59, 64, 67, 70–71, 74, 76, 78, 81, 82) | 68.82 | 1.273 (1.152–1.405) | 0.000 | 0.000 | 71.4% | |
| Study score | 0.658 | ||||||
| Low | 14 (32, 35, 37, 39, 43, 54, 60–62, 65, 68, 72, 77) | 30.37 | 1.434 (1.133–1.814) | 0.003 | 0.000 | 89.7% | |
| High | 35 (22, 26–27, 29–30, 36, 38, 40–42, 44–47, 49–51, 53, 55–59, 64, 67, 70–71, 74, 76, 78, 81, 82) | 69.63 | 1.273 (1.152–1.407) | 0.000 | 0.000 | 71.3% | |
| Exposure assessment | 0.077 | ||||||
| Database | 10 (32, 35, 39, 60–62, 65, 68, 77) | 21.85 | 1.452 (1.211–1.741) | 0.000 | 0.001 | 67.2% | |
| Questionnaire | 12 (22, 29, 36, 45, 47, 59, 70, 74, 76, 81, 82) | 21.42 | 1.380 (1.137–1.676) | 0.001 | 0.001 | 65.2% | |
| Interview | 25 (26–27, 30, 37–38, 40–44, 46, 49–51, 53, 55–57, 64, 67, 71–72, 78) | 49.54 | 1.287 (1.126–1.471) | 0.000 | 0.000 | 68.6% | |
| Other | 2 (54, 58) | 7.19 | 0.906 (0.596–1.378) | 0.645 | 0.000 | 97.6% | |
| Occupation | 0.350 | ||||||
| Nurses | 11 (29, 36, 38, 39, 42, 65, 67, 68, 82) | 21.02 | 1.577 (1.235–2.014) | 0.000 | 0.000 | 72.0% | |
| Unclassified occupation | 30 (22, 26–27, 30, 32, 35, 37, 40–41, 43–46, 49–51, 53, 55–58, 64, 70–72, 74, 76, 78, 81) | 63.60 | 1.250 (1.130–1.383) | 0.000 | 0.000 | 73.3% | |
| Flight attendants | 5 (59–62) | 7.98 | 1.454 (1.100–1.922) | 0.009 | 0.161 | 39.0% | |
| Military women | 1 (47) | 1.13 | 2.100 (0.990–4.455) | 0.053 | — | — | |
| Textile workers | 1 (54) | 3.55 | 0.730 (0.655–0.814) | 0.000 | — | — | |
| Radio and telegraph operators | 1 (77) | 2.73 | 1.500 (1.112–2.023) | 0.008 | — | — | |
aSome enrolled articles discussed more than one type of cancer separately, when calculating the number of studies; we took different cancers into consideration in one article.
Dose–response meta-analysis
Dose–response meta-analysis was performed among studies that involved at least 3 levels of exposure categories. The number of studies on breast cancer accounted for an overwhelming proportion and sixteen studies were included to quantitatively assess the cumulative effect of exposure to night shift work on breast cancer incidence. As for other types of cancers, the number of relevant literature was not enough for dose–response meta-analysis. For every 5 years of night shift work, the risk of breast cancer in women increased by 3.3% (OR = 1.033; 95% CI, 1.012–1.056) as shown in Fig. 3.
OR of breast cancer in women by years of night shift work based on dose-response meta-analysis. Solid line represents the estimated OR and the dotted lines represent the low limit and upper limit of 95% CIs.
OR of breast cancer in women by years of night shift work based on dose-response meta-analysis. Solid line represents the estimated OR and the dotted lines represent the low limit and upper limit of 95% CIs.
Study quality
The NOS was employed for quality evaluation of eligible studies. Their scores ranged from 5 to 8 and the mean value was 7.1, indicating a favorable overall quality. Among the included studies, 43 articles were considered to be of high quality with scores ≥7 and 8 studies got 5 scores because of lack of complete research records and variables of adjustment.
Publication bias
We performed Begg funnel plot to assess the publication bias of included studies. Potential publication bias was identified (P = 0.006) among all retrieved studies probably due to the variety of involved cancers, whereas no publication bias was observed among studies on breast cancer (P = 0.208; Supplementary Fig. S2).
Discussion
Some people in contemporary society work in a 24-hour mode, disrupting the 8-hour day routine (83). With the productive and economic development, night shift work is strongly required in the fields of industry, commerce, and entertainment. Shift workers suffer from disturbance of circadian rhythm and suppression of nocturnal melatonin. Short-term effects of night shift work were summarized as “jet-lag” syndrome, including sleep disorders, digestive troubles, fatigue, emotional fluctuation, and reduced physical activity. Long-term night shift work was reported to be associated with increased risks of cardiovascular disease, neuropsychiatric disorder, endocrine system disorders, and cancer (84–87). Data from the third EU Survey (2000) showed that 76% employee worked beyond normal working time (88). Up to 21.9% of men and 10.7% of women were exposed to shift work, with 7% population working permanently at night (89). Large numbers of people are being exposed to night shift work, which brings huge detrimental impact on health; it is therefore of much significance to conduct the study to illustrate the relationship between night shift work and the risks of frequently-occurred cancers in women.
Sixty-one studies were enrolled in the meta-analysis including 26 cohort studies, 24 case–control studies, and 11 nested case–control studies with 114,628 cases and 3,909,152 participants from Europe, North America, Asia, and Australia. A positive dose–response relationship was present between night shift work and the risks of breast cancer (OR = 1.316; 95% CI, 1.196–1.448), digestive system cancer (OR = 1.177; 95% CI, 1.065–1.301), and skin cancer (OR = 1.408; 95% CI, 1.024–1.934). Among the group of nurses, long-term night shift work increased the risks of breast cancer (OR = 1.577; 95% CI, 1.235–2.014), digestive system cancer (OR = 1.350; 95% CI, 1.030–1.770), and lung cancer (OR = 1.280; 95% CI, 1.070–1.531). A nonsignificant effect was observed for ovarian cancer (OR = 1.135; 0.970–1.328) and no effect was seen for cervical cancer (OR = 0.980; 95% CI, 0.787–1.221). Night shift work elevated the risk of breast cancer in a dose–response way which was consistent with previous studies (6, 20, 90). For every 5 years of night shift work, the risk of breast cancer in women increased by 3.3% (OR = 1.033; 95% CI, 1.012–1.056).
Night shift work causes an increase in sex hormones, which is speculated to be relevant for hormone-dependent cancers (91). Strong epidemiologic evidence supports the association between night shift work and increased risk of breast cancer, and also there is limited evidence on prostate cancer (92, 93) and endometrial cancer (25). In our analysis, the risks of hormone-sensitive cancers including breast cancer, ovarian cancer, and uterine cancer among night shift workers were shown in Table 2. Night shift work was strongly associated with higher risk of breast cancer in females whereas no effect was observed for ovarian cancer and uterine cancer.
The underlying biological mechanisms of the association between night shift work and increased cancer risks are complex. One of the possible hypotheses is that exposure to light at night accompanying night shift work results in the disruption of circadian rhythm and the reduction of melatonin production (5). Melatonin is characterized with oncostatic effect which works through antioxidation, antiangiogenesis, regulation of immunity, and metabolism (7). Melatonin reduction stimulates the production of pituitary gonadotropins by negative feedback, hence increasing the risk of sensitive cancers such as breast cancer, ovarian cancer, and endometrial cancer (8). Animal experiments have demonstrated that long-term oral melatonin supplement offered a protective effect against breast cancer (94). Several in vitro researches showed that melatonin administration of even biological dose had a significant growth inhibition effect on breast cancer cells (95–98) and other tumor cells (99–103). Sun exposure is sharply decreased among night shift workers, leading to reduced vitamin D level. Some experimental and epidemiological researches supported the inverse correlation between circulating vitamin D and the risk of breast cancer (104, 105) or colorectal cancer (106, 107). However, night shift work was often accompanied with irregular eating habits, which somewhat contributed to digestive system tumors. Results in our study were inconsistent with the above theories that night shift work increased the risk of hormone-sensitive cancers and digestive system cancers.
Q and I2 statistics were used to evaluate heterogeneity. As a result, significant heterogeneity was observed in the groups of breast cancer (P = 0.000, I2 = 80.4%), skin cancer (P = 0.009, I2 = 64.7%), and uterine cancer (P = 0.042, I2 = 59.6%). Random-effect model was adopted in an attempt to eliminate all sources of heterogeneity. Furthermore, subgroup analyses about breast cancer were conducted and less heterogeneity existed among cohort studies, high-score studies and studies adjusted for >3 confounders, suggesting that these studies would provide more reliable evidence. Moreover, we carried out meta-regression analysis and no explanation was found for possible heterogeneity sources from variables due to low statistical power, therefore the results of risk estimates should be interpreted with caution. When stratified by region, positive correlation existed in Europe and North America, but not in Asia and Australia. One possible reason might be that Asian population was less sensitive to nightshift exposure. Another was partly attributed to the difference of sleeping habits, economic development, and medical service across different geographical areas. Short sleep duration, light at night, and airline cabin crew servings involved potential circadian rhythm disturbances just like nightshift; therefore, relevant articles were included in this meta-analysis to reduce selection bias. Flight attendants were simultaneously exposed to cosmic radiation which was a potential cancer-related unmeasurable factor, thus the odds risk might be overestimated. Among nurses, remarkable elevation of cancer risk was observed regarding breast cancer (OR = 1.577; 95% CI, 1.235–2.014), digestive system cancer (OR = 1.350; 95% CI, 1.030–1.770), and lung cancer (OR = 1.280; 95% CI, 1.070–1.531). In the future, large-sample and multiregion researches are needed to update and confirm the association.
The present meta-analysis involved 3,909,152 participants and 114,628 female patients with cancers at 11 sites. To the best of our knowledge, this is the first meta-analysis to comprehensively assess the association between night shift work and the risk of common cancers among female population. Compared with previous meta-analyses, the study has the following merits. First, in the methodological aspect we systematically conducted risk estimates, subgroup analyses, meta-regression analysis, and dose–response meta-analysis. Random effect model was used to eliminate the source of heterogeneity to some extent. In addition, subgroup analyses and meta-regression were performed to explore potential sources of heterogeneity from confounding factors. Second, according to inclusion and exclusion criteria, 61 studies were enrolled in the meta-analysis and the accumulated evidence with enlarged sample size enhanced statistical power to derive a more precise and reliable risk estimation. Beyond that, longest versus shortest duration was taken as exposure indicator and each individual article was involved in the pooled risk estimate, increasing the generalizability of results. Third, the majority of included studies (40 out of 67) were carried out among general population, therefore the estimation of association between night shift work and cancer risk of women could be extended, not limited to some particular working groups. A certain amount of studies (17 out of 67) were based on nursing group which was an important part of female shift workers and a stratified analysis revealed that night shift work increased the risk of breast cancer, lung cancer and digestive system cancer in female nurses. Fourth, cohort study is less susceptible to confounding factors and less affected by recall bias, thus result from cohort study is considered more credible and valuable compared with case–control study. In this meta-analysis, 26 cohort studies with enough follow-up period were included and a positive relationship was found in the cohort group.
Nevertheless, some shortcomings in the study have to be mentioned. First, the period time of night shift across all enrolled studies was not defined uniformly, for example one definition was “working at least three nights per month” (29) and another was “working during the night (23:00–06:00 h)” (28). Included studies used very different definitions of period time regarding night shift and the lack of consistent definition might result in a certain degree of misclassification, consequently leading to a dilution of pooled estimates when doing data synthesis. Second, a significant between-study heterogeneity and publication bias was observed. Significant variability existed in different individual studies regarding study population, geographical location, adjustment confounders, study design and exposure definition, and each of these factors may contribute to heterogeneity. Disappointingly, we failed to find out the possible sources of heterogeneity from a meta-regression due to the low statistical power. Given substantial heterogeneity observed among included studies, the evidence supporting the association might be weakened and additional well-designed researches are needed. Third, case–control study are vulnerable to varying levels of bias, and in general patients in case group tend to recall past exposure history of night shift. Twenty-four case–control studies were included in our analysis which could probably bring about selection and recall bias. Also, unstandardized questionnaires might produce information bias and inadequate control of adjustment factors could lead to either underestimation or exaggeration of the pooled risk. Finally, only a small number of enrolled studies were involved in dose-response meta-analysis, hence limiting the reliability of the result.
In conclusion, this meta-analysis updated previous studies and identified that cancer risk in female population was increased with accumulating years of night shift work. Given the expanding prevalence of shift work worldwide and heavy public burden of cancers, further researches, particularly large-size, high-score cohort studies are of great necessity to confirm the relationship between night shift work and cancer risk. Also, in-depth biological researches should be done to explore the mechanisms by which night shift work affects cancer risk. Knowing how night shift work serves as a risk factor for cancers might help establish and implement effective measures to protect female night shifters. Moreover, it is important that long-term night shift workers accept regular physical examination and cancer screening for potential malignancies, particularly breast cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
RSS Feeds