The incidence of breast cancer among female flight attendants: an updated meta-analysis

Transcription

1 Journal of Travel Medicine, 2016, 1 7 doi: /jtm/taw055 Review Review The incidence of breast cancer among female flight attendants: an updated meta-analysis Tiebing Liu 1,, *, Chanyuan Zhang 2,, and Chong Liu 3 1 Civil Aviation Medicine Center, Civil Aviation Administration of China, Beijing, People s Republic of China, 2 Department of Clinical Laboratory, Civil Aviation General Hospital, Beijing, People s Republic of China and 3 Department of Information Engineering, Cangzhou Technical College, Cangzhou, Hebei, People s Republic of China *To whom corresponding should be addressed. ltbing@yeah.net These authors contributed equally to this work. Submitted 2 June 2016; revised 23 July 2016; Accepted 1 August 2016 Abstract Background: Several studies have indicated an increased risk of breast cancer (BC) among female flight attendants (FFAs); however, the results from epidemiological studies were not consistent. We thus conducted an updated meta-analysis to re-assess the risk of BC among FFAs, according to the MOOSE guideline. Methods: A systematical search of PubMed and Embase for relevant observational studies up to March 2016 was performed, supplemented by manual reviews of bibliographies in relevant studies. A random effect model was conducted to calculate the combined standard incidence ratio (SIR) and 95% confidence interval (95% CI) in BC risk. Results: Of the 719 citations retrieved, 10 were included, with more than participants and 821 new cases. The combined SIR (95% CI) for BC in FFAs was 1.40 (95%CI ), with no significant heterogeneity (P ¼ 0.744; I 2 ¼ 0.0%) or publication bias (Begg s test: z ¼ 0.72, P ¼ 0.474; Egger s test: t ¼ 0.25, P ¼ 0.805) among the included studies. The results were not significantly modified by publication year, geographic area, study quality or whether the fertility variables were adjusted. Conclusions: Our meta-analysis suggests that FFAs have a higher risk of BC compared with the general population. More vigorous studies with larger sample sizes based on other populations, including the Chinese, are needed. Key words: Breast cancer incidence, female flight attendants, meta-analysis, cosmic radiation, circadian disruption Introduction Breast cancer (BC), the most frequently diagnosed cancer, is also the major cause of cancer death in women worldwide in 2012, with 1.7 million cases and deaths. 1 3 Flightbased workers are thought to have a greater occupational hazard risk of BC owing to increased altitude-related exposure to cosmic radiation and disrupted sleep patterns. However, epidemiological studies regarding BC risk in female flight attendants (FFAs) compared with the general population have yielded inconsistent results. Several studies have reported significantly higher incidences of BC in FFAs, 4 6 whereas others did not observe significant different BC incidence between FFAs and the general public Both of the two earlier meta-analyses, including relatively small number of studies with broad confidence intervals (CI) of standard incidence ratio (SIR), were both published in 2006 and took the incidence of numerous cancer types into consideration including BC. 12,13 Recently, there were three more cohort studies indicating inconsistent results. 4,5,11 Therefore, we carried out a meta-analysis to update and quantitatively re-assess the BC risk among FFAs on a larger sample size by combining the conflicting results of all available studies. Thus, we aimed at increasing the precision of risk estimates of BC incidence among FFAs. Methods Ethical approve is not required for our meta-analysis. This study was performed according to the meta-analysis of observational studies in epidemiology criteria (MOOSE). 14 VC International Society of Travel Medicine, Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oup.com

2 2 Journal of Travel Medicine, 2016, Vol. 23, No. 6 Search Strategy Two authors (T.L. and C.Z.) independently searched PubMed and Embase for epidemiological observational studies up to March The authors of potential publications were contacted when we needed more data. Two groups of keywords were used in the searching strategy: (1) breast cancer, breast carcinoma, breast neoplasm ; (2) flight attendant, flight personnel, cabin crew, aircrew, cabin attendant. We also performed manual searches of the reference lists of the retrieved studies and reviews to identify additional eligible studies. Selection Criteria and Study Selection Publications included in this meta-analysis should adhere to the following inclusion criteria: (1) evaluation of BC incidence among FFAs; (2) standardized incidence ratio (SIR) with standard errors (SE) or confidence intervals (CI); (3) no language restriction. SIR was a measure of the incidence in a study population (in this study, FFAs) in comparison with the general population. SIR was normally standardized by age, sex and calendar year. Figures for SIR greater than 1 suggested a higher incidence among the study population when compared with the general population (a national or state reference population). Two authors (T.L. and C.Z.) independently assessed the literature eligibility and conflict was resolved by a third author (C.L.). Quality Assessment and Data Extraction Two authors (T.L. and C.Z.) assessed aspects of the quality of each selected study with the Newcastle-Ottawa quality scale (NOS), 15 a validated scale for non-randomized studies in metaanalysis. This scale awards a maximum of nine points to each study: (1) four points for adequate selection of participants, (2) two points for comparability for cases and controls based on the design or analysis and (3) three points for the adequate assessment of outcomes. A NOS score (6 points) was defined as a high-quality study. Discrepancy in quality assessment was resolved by a third author (C.L.). Two authors (C.Z. and C.L.) independently extracted the data with a predesigned data extraction form. The following information were extracted: (1) first author s name; (2) year of publication; (3) location or country where the study was performed; (4) number of BC cases; (5) follow-up person-years; (6) study period; (7) SIR with 95%CI; (8) confounding factors adjusted in the analysis. Discrepancies in data extraction were resolved by a third author (T.L.). Data Analysis For the primary analyses, the SIR of BC among FFAs were summarized. In spite of whether heterogeneity existed between studies, a random effect model was performed to evaluate combined SIR with 95% CI due to the small number of included studies. Between-study heterogeneity was determined by the Q test and I 2 statistic. The statistical significance for the Q test was defined as P < The I 2 statistic was calculated to represent the size of total variation accounted by heterogeneity. I 2 values of 25, 50 and 75% indicated low, moderate and high degrees of heterogeneity, respectively. To identify the stability of the primary results and to examine the resource of potential heterogeneity, we performed subgroup analyses by publication year (before 2010 or after 2010), region where the study was conducted (USA or Europe), whether the fertility variables were adjusted (yes or no), and study quality (6 or <6). For sensitivity analyses, leave-one-out analyses were used to investigate the magnitude of influence of each study on pooled risk estimates. 16 We evaluated the possibility of publication bias by the Egger s test, 17 Begg s test 18 and visual inspection with a funnel plot. The statistical analyses were conducted using Stata, statistical software (version 12.0, StataCorp, USA). Statistical tests were twosided and P < 0.05 was considered statistically significant. Results Literature Search Figure 1 demonstrated the detailed literature search strategy for this meta-analysis. We retrieved 718 studies from the Pubmed and Embase database, and 1 from the reference list of relevant studies. After 58 duplicate studies were excluded, we screened 661 studies through titles and abstracts, of which 614 were excluded because they did not meet the inclusion criteria. After reviewing the full text of the remaining 47 potentially eligible studies carefully, 37 were excluded because they were reviews, cross-sectional studies or irrelevant studies. Finally, a total of 10 studies were included in the meta-analysis. Among the included studies, one was retrieved from the reference list. 19 Study Characteristics Characteristics of the 10 studies included in the meta-analysis were shown in Table 1. The studies included in this metaanalysis were published between 1995 and 2015, reported data from 1947 to 1997, and included more than participants, with 821 new cases and more than follow-up person-years. Three studies were from the USA, and seven from Europe. The duration of follow-up ranged from 7 to 50 years. The quality scores were from 3 to 8. The study by Pukkala et al. 4 was a pooled analysis including data from four countries (Finland, Iceland, Norway and Sweden). Main Analysis Results from the 10 studies were combined by a random effect model to provide information for BC incidence among FFAs. Compared with the general population, the combined SIR (95% CI) for BC among FFAs was 1.40 (95%CI ; 10 records) (Figure 2). Heterogeneity was not significant across the included studies (P ¼ 0.744; I 2 ¼0.0%). Sensitivity Analysis and Publication Bias To examine the stability of the primary results, leave-one-out analyses were conducted to explore the magnitude of influence of each study on the pooled risk estimates, with a range of 1.38 (95%CI ) to 1.45 (95%CI ), indicating that the overall risk estimates were not substantially modified by any single study. Visual inspection of a funnel plot did not reveal asymmetry (Figure 3). The Egger s test and Begg s test also suggested no

3 Journal of Travel Medicine, 2016, Vol. 23, No. 6 3 Figure 1. Flowchart of study identification and inclusion evidence of obvious publication bias across the studies (Begg s test: z ¼ 0.72, P ¼ 0.474; Egger s test: t ¼ 0.25, P ¼ 0.805). Subgroup Analysis According to publication year (before 2010 or after 2010), geographic area (USA or Europe), study quality (6 or <6) and whether the fertility variables were adjusted (yes or no), we conducted subgroup analyses with random-effect models to identify the stability of the primary results and to examine the source of potential heterogeneity. An increased incidence of BC among FFAs compared with the general population and no evidence of heterogeneity was observed among all the subgroup analyses (all P >0.10). When the analysis was stratified by geographic area, the pooled SIR estimate obtained among studies in Europe (SIR ¼ 1.42; 95%CI ; P for heterogeneity ¼ 0.518; I 2 ¼ 0.0%) was higher than that among studies in USA (SIR ¼ 1.38; 95%CI ; P for heterogeneity ¼ 0.739; I 2 ¼ 0.0%). When stratifying studies by publication year, the pooled SIR was 1.35 (95%CI ; P for heterogeneity ¼ 0.671; I 2 ¼0.0%) for studies published before 2010, and 1.41 (95%CI ; P for heterogeneity ¼ 0.278; I 2 ¼ 0.0%) for studies published after When subgroup analysis were stratified by quality score, we found that studies with a lower quality score tended to have a higher BC incidence (SIR ¼ 1.42; 95%CI ; P for heterogeneity ¼ 0.920; I 2 ¼ 0.0%) as compared with studies with a higher quality score (SIR ¼ 1.39; 95%CI ; P for heterogeneity ¼ 0.286; I 2 ¼ 20.2%). When the analysis was stratified by whether controlling for fertility variables, the pooled SIR controlling for fertility variables was lower than that without fertility variables in the models [(SIR¼ 1.39; 95%CI ; P for heterogeneity ¼ 0.320; I 2 ¼14.5%) vs (SIR ¼ 1.42; 95%CI ; P for heterogeneity ¼ 0.788; I 2 ¼0.0%)]. However, the results of subgroup analyses were not significantly different due to the overlapping confidence intervals; therefore, the combined SIR of BC was not significantly modified by publication year, geographic area, study quality or whether the fertility variables were adjusted. The results of subgroup analyses were presented in Table 2. Discussion Travelling is becoming increasingly widespread in modern world, and the health issues of FFAs have been of increasing interest to the civil aviation health researchers. This meta-analysis incorporated 10 existing published observational studies, including 821 new cases and participants with more than half of a million person-years of follow-up, and provided a quantitatively risk estimate of BC in FFAs. Compared with the general population, we observed that the combined SIR for FFAs was 1.40 (95%CI ), which indicated that FFAs

4 4 Journal of Travel Medicine, 2016, Vol. 23, No. 6 Table 1. Main characteristics of studies included in the meta-analysis Authors Country Study population Publication year New cases No. of Person-years Years of workers a follow-up NOS Confounders adjusted SIR (95%CI) Pukkala et al. 20 Finland FFAs of Finnish airline companies, Finland Lynge et al. 9 Denmark Danish female airline FA registered in the 1970 census Wartenberg and Stapleton 19 USA US retired female FA of an airline company Rafnsson et al. 10 Iceland FFAs from the Iceland Cabin Crew Association Age, sex, calendar year 1.87 ( ) Not stated Age, sex, calendar year 1.61 ( ) Not stated Not stated 3 Age, sex, calendar year 2.00( ) Age, sex, calendar year, age at first birth, parity, number of children Haldorsen et al. 7 Norway licensed FA in Norway Age, sex, calendar year, age Reynolds et al. 6 USA Members of Association of Flight Attendants with California residence Linnersjö et al. 8 Sweden FFAs of Swedish Pukkala et al. 4 Finland, Iceland, Norway, Sweden Scandinavian Airline System Various registers in Finland, Iceland, Norway and Sweden dos Santos Silva et al. 11 UK FFAs from the Medical Records System (MRS) of the UK Civil Aviation Authority (CAA) Schubauer-Berigan et al. 5 USA FFAs from the personnel records of Pan American World Airways (Pan Am) at first birth, length of employment Not stated Age, sex, calendar year, flight assignment, years of service, age at entry 1.45 ( ) 1.12 ( ) 1.42 ( ) Age, sex, calendar year 1.3 ( ) b 8 Age, sex, calendar year, age at first birth, parity 1.5 ( ) Not stated Age, sex, calendar year 1.13 ( ) Age, sex, calendar year, age at first birth, parity 1.37 ( ) a Female. b Variable per country.

5 Journal of Travel Medicine, 2016, Vol. 23, No. 6 5 Figure 2. Meta-analysis of observational studies on BC incidence among FFAs Table 2. Subgroup analyses of the SIR of BC in FFAs compared with the general population Study group No. of studies SIR (95%CI) P for heterogeneity I 2 (%) Figure 3. Funnel plot for publication bias had a moderately increased incidence of 40% for BC. In addition, there was no evidence of substantial heterogeneity and obvious publication bias across the included studies in our meta-analysis. Of the 10 studies examined, the majority indicated an increased incidence of BC among FFAs compared with the general population; however five reported conflicting results, 7 11 which did not demonstrate significantly different incidence of BC between FFAs and the general public. The pooled results of our meta-analysis were consistent with several studies 4 6,19,20 analysed and two earlier meta-analysis 12,13 [SIR ¼ 1.40 (95%CI ) and 1.41 (95%CI ), respectively], and an increased risk of BC among FFAs was noted. Moreover, we observed that all the pooled SIR estimate points were higher than 1 in all subgroup analyses. In addition, the range of corresponding 95% CI became narrower when the number of studies and sample size increased. Our results suggested that 40% of BC in these FFAs could be possibly attributable to their occupational exposure to risk All ( ) Geographic area Europe ( ) USA ( ) Publication year Before ( ) After ( ) NOS ( ) < ( ) Controlling for fertility variables No ( ) Yes ( ) factors, mainly including cosmic radiation and circadian disruption. The dose of occupational exposure to cosmic radiation for FFAs was twice that for the general population, 4,21 although it was still less than the allowed dose limit [20 millisieverts (msv) per year] In addition, the levels of cosmic radiation exposure increased along with longer flights, more solar activities and higher altitudes, especially in the polar areas. 8,25 Although the Federal Aviation Administration (FAA) offers good estimates about the levels of cosmic radiation exposure for crewmembers; doses from solar proton events are still overlooked. Furthermore, the radiation dose created by a thunderstorm was estimated at 30 msv, which was higher than the allowed dose limit (20 msv per year). 26 Moreover, flight histories of annual average dose estimates for cabin crew have not been recorded as precisely as flight pilots. 27 Considering the ability of cosmic radiation to damage DNA in cells, it might be associated with an increased risk of BC among FFAs.

6 6 Journal of Travel Medicine, 2016, Vol. 23, No. 6 The development of BC has also been linked to the disruption of circadian rhythms, which is usually caused by shift work, short sleep duration and exposure to light at night. 28, 29 The underlying mechanism is related to the increased exposure to the hormonal alteration, which could be resulted by the circadian disturbance. 30 In addition, circadian disruption was reported to lead to significant changes in immune system and biological processes associated with BC risk. 31 For FFAs, the potential causes for the circadian rhythm disruption mainly include number of hours spent flying during the standard sleep interval and number of time zones crossed. In animal study of mice, simulated chronic jet lag, which was linked to crossing multiple time zones, could result in circadian disruption and cause tumour growth significantly. 32 Furthermore, a significant link between disruption of menstrual cycle owing to jet lag and BC risk was observed in a case control analysis regarding Finnish cabin crew. 30 Similarly, a meta-analysis showed an significant association between BC risk in women and circadian disrupting exposures when aggregating all studies regardless of the source of circadian disruption (RR ¼ 1.14, 95%CI ). 28 It is also possible that non-occupational factors could contribute to the increased risk of BC among FFAs. This was supported by a nested case control study which did not show any association between the occupational factors (including sleep rhythm disruptions, disruption of menstrual cycles and cumulative radiation does) and BC risk among Finnish flight attendants. In addition, previous studies in Finland, Iceland and Germany suggested that FFAs were more likely to have several reproductive risk factors for BC than the general population. 10,20 However, known non-occupational risk factors for BC including parity, age at first birth, family history of BC and socioeconomic status have been suggested to account for only a fraction of cases, 6,10,33 which means that these observed differences were insufficient to explain the magnitude of excess observed. Similarly, in the subgroup analysis, when the analysis was stratified by whether controlling for fertility variables, the pooled SIR was higher when controlling for fertility variables compared with that without fertility variables in the models, indicating 0.3% of BC could be attributed to fertility factors. Our meta-analysis of 10 studies involving participants and 821 new cases improved the statistical power and found a more reliable risk estimates of BC in FFAs. All investigations were cohort studies, which could reduce the recall and selection bias. In addition, cancer incidence is a better risk indicator that mortality for tumours, like BC, with a high chance of recovery in case of early diagnosis. Moreover, there was no evidence of substantial heterogeneity and obvious publication bias across the included studies in our meta-analysis. However, several limitations in the present meta-analysis should be considered. First, several other factors may influence the overall risk estimates of BC in FFAs. Although most studies included studies were adjusted for a wide range of factors, some did not exclude the influence of potential factors. Second, regarding the general population as a reference might lead to underestimation of the risk of cancer among working populations because of the healthy worker effect, which presumably derived from screening processes, allowing relatively healthier people to become or remain workers. 34,35 Third, the study populations varied among the included studies. Specifically, three studies were conducted in the USA and seven in Europe. Although there was no indication of statistical heterogeneity, we could not exclude the disturbing impact of potential clinical heterogeneity, due to the fact that the overall estimate of this meta-analysis was based on only 10 studies. In addition, the chance of publication bias was still an intractable issue for this meta-analysis of limited number of qualified studies. Owing to these limitations, the interpretation and extrapolation of this meta-analysis were restricted. Therefore, more vigorous studies with larger sample sizes based on other populations, including the Chinese, are needed. Conclusion The findings of this meta-analysis suggested that FFAs had a significantly increased risk of BC compared to the general population. This has important implications for occupational health and protection of the population of FFAs. However, further prospective-designed studies with more ethnic groups and larger sample size are needed to validate the risk observed in the present meta-analysis. Conflict of interest: None declared. References 1. Torre LA, Bray F, Siegel RL, et al. Global cancer statistics, CA Cancer J Clin 2015; 65: Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in 2008: GLOBOCAN Int J Cancer 2010; 127: Desantis C, Ma J, Bryan L, et al. Breast cancer statistics, CA Cancer J Clin 2014; 64: Pukkala E, Helminen M, Haldorsen T, et al. Cancer incidence among Nordic airline cabin crew. Int J Cancer 2012; 131: Schubauer-Berigan MK, Anderson JL, Hein MJ, et al. Breast cancer incidence in a cohort of U.S. flight attendants. Am J Ind Med 2015; 58: Reynolds P, Cone J, Layefsky M et al. Cancer incidence in California flight attendants (United States). Cancer Causes Control 2002; 13: Haldorsen T, Reitan JB, Tveten U. Cancer incidence among Norwegian airline cabin attendants. Int J Epidemiol 2001; 30: Linnersjo A, Hammar N, Dammstrom BG, et al. Cancer incidence in airline cabin crew: experience from Sweden. Occup Environ Med 2003; 60: Lynge E. Risk of breast cancer is also increased among Danish female airline cabin attendants. BMJ 1996; 312: Rafnsson V, Tulinius H, Jonasson JG, et al. Risk of breast cancer in female flight attendants: a population-based study (Iceland). Cancer Causes Control 2001; 12: dos Santos Silva I, de Stavola B, Pizzi C, et al. Cancer incidence in professional flight crew and air traffic control officers: disentangling the effect of occupational versus lifestyle exposures. Int J Cancer 2013; 132: Tokumaru O, Haruki K, Bacal K, et al. Incidence of cancer among female flight attendants: a meta-analysis. J Travel Med 2006; 13: Buja A, Mastrangelo G, Perissinotto E, et al. Cancer incidence among female flight attendants: a meta-analysis of published data. J Womens Health (Larchmt) 2006; 15:

7 Journal of Travel Medicine, 2016, Vol. 23, No Stroup DF, Berlin JA, Morton SC. et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Metaanalysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000; 283: Wells GA, Shea B, O Connell D, et al. The Newcastle Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. (23 January 2016, date last accessed). 16. Wallace BC, Small K, Brodley CE, et al. Toward modernizing the systematic review pipeline in genetics: efficient updating via data mining. Genet Med 2012; 14: Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994; Wartenberg D, Stapleton CP. Risk of breast cancer is also increased among retired US female airline cabin attendants. BMJ1998; 316: Pukkala E, Auvinen A, Wahlberg G. Incidence of cancer among Finnish airline cabin attendants, BMJ 1995; 311: Bartlett DT. Radiation protection aspects of the cosmic radiation exposure of aircraft crew. Radiat Prot Dosimetry 2004; 109: Zeeb H, Hammer GP, Blettner M. Epidemiological investigations of aircrew: an occupational group with low-level cosmic radiation exposure. J Radiol Prot 2012; 32: N Grajewski B, Waters MA, Yong LC, et al. Airline pilot cosmic radiation and circadian disruption exposure assessment from logbooks and company records. Ann Occup Hyg 2011; 55: Waters M, Bloom TF, Grajewski B. The NIOSH/FAA Working Women s Health Study: evaluation of the cosmic-radiation exposures of flight attendants. Federal Aviation Administration. Health Phys 2000; 79: Whelan EA. Cancer incidence in airline cabin crew. Occup Environ Med 2003; 60: Bramlitt ET, Shonka JJ. Radiation exposure of aviation crewmembers and cancer. Health Phys 2015; 108: Pukkala E, Aspholm R, Auvinen A, et al. Incidence of cancer among Nordic airline pilots over five decades: occupational cohort study. BMJ 2002; 325: He C, Anand ST, Ebell MH, et al. Circadian disrupting exposures and breast cancer risk: a meta-analysis. Int Arch Occup Environ Health 2015; 88: Weiderpass E, Meo M, Vainio H. Risk factors for breast cancer, including occupational exposures. Saf Health Work 2011; 2: Kojo K, Pukkala E, Auvinen A. Breast cancer risk among Finnish cabin attendants: a nested case-control study. Occup Environ Med 2005; 62: Blask DE, Hill SM, Dauchy RT, et al. Circadian regulation of molecular, dietary, and metabolic signaling mechanisms of human breast cancer growth by the nocturnal melatonin signal and the consequences of its disruption by light at night. J Pineal Res 2011; 51: Filipski E, Levi F. Circadian disruption in experimental cancer processes. Integr Cancer Ther 2009; 8: Winter M, Blettner M, Zeeb H. Prevalence of risk factors for breast cancer in German airline cabin crew: a cross-sectional study. J Occup Med Toxicol 2014; 9: Wagner LK. The Healthy Worker Effect : science or prejudice? Radiology 2003; 229: Kirkeleit J, Riise T, Bjorge T, et al. The healthy worker effect in cancer incidence studies. Am J Epidemiol 2013; 177: