To the best of our knowledge, other than Saito et al. , who carried out a cross-sectional study (n=105) that investigated work-related as opposed to clinical factors (e.g., cancer stage, surgery), this is the first study to investigate predictors of job resignation and SL among BCSs in Japan. We found that 14.9% of the BCSs in this study quit their jobs at least 1 year after being diagnosed with breast cancer. In addition, the post-cancer diagnosis resignation rate differed significantly according to education level, cancer stage, and occupational type. A systematic review reported that CSs were more likely to be unemployed than were healthy controls (33.8% vs. 15.2%, respectively; pooled relative risk: 1.37) , which suggests that developed countries support CSs to avoid potentially high numbers of resignations . The resignation rate (14.9%) of BCSs in this study was lower than that reported in the previous systematic review . Endo et al.  reported that resignation rates were quite low among total cancer in Japan (12.4%), where it is very difficult and uncommon for employers to fire employees. The Labor Contract Act of Japan states the following: “A dismissal shall, if it lacks objectively reasonable grounds and is not considered to be appropriate in general societal terms, be treated as an abuse of right and be invalid” .
This study found that age at diagnosis, lower education level, and taking SL were predictors of resignation after breast cancer diagnosis; predictors of taking SL were limited to having undergone surgery. We therefore speculated that being highly educated or taking SL might be confounded by being able to access the SL scheme for workers at larger companies easily, as the SL system is better established in larger than in smaller companies . Since the results from this study might depend on the availability of SL, the relationship between the length of SL or the work environment and resignation after breast cancer diagnosis should be studied in the future.
Regarding predictors of resignation after breast cancer diagnosis, first, our findings indicated that younger BCSs resigned more frequently than their older counterparts, in accordance with previous studies that argue that young BCSs have a higher risk of losing paid employment because breast cancer and its associated treatment are often more aggressive at a younger age, suggesting that young BCSs may experience more severe long-term adverse effects, including those that are work-related (or related to substance of work) [22, 23]. In addition, older people may have more knowledge and technology related to the companies and work compared with younger people . Our data suggest that older BCSs may be more reticent to resign, given the typical age-associated difficulties in finding new employment. However, Fantoni et al.  reported that older age was associated with difficulty continuing work and a higher risk of unemployment. Further studies exploring the reasons behind resignation are therefore warranted.
Second, patients with lower compared with higher educational attainment were found to be at higher risk for resignation. This finding is consistent with previous studies of non-Asian populations [12, 24, 26–28]. However, a comparison of resignation rates with studies from other countries warrants careful consideration, given the important differences in socioenvironmental factors, including the widely differing regulation of medical leave provision by national systems and the availability of company-based health care resources . In addition, income has been shown to be correlated with education level: lower income has been found to be associated with an increased likelihood of resignation and unemployment among BCSs [12, 24, 30–32]. Furthermore, educational attainment is likely related to occupation type, with less educated individuals more likely to be working in physically demanding jobs such as manual labor . A MHLW survey in Japan found that people with lower education levels were more likely to have physically demanding jobs such those in the hospitality and wholesale and retail trade industries . Employees with more physically demanding jobs such as manual labor and blue-collar work are more susceptible to resignation [12, 24, 28, 35, 36]. Petersson et al. [37, 38] reported that higher education level was related to greater dedication to work, and that RTW was earlier in patients who valued their work more highly.
Third, our results indicated that the risk of resignation was substantially higher among BCSs who took SL after breast cancer diagnosis than among those who did not. These findings are consistent with previous studies that showed a correlation between length of SL and RTW, with longer SL making RTW and continued employment more difficult [39, 40]. Conversely, Azarkish et al.  found no relationship between taking SL and job loss. Longer SL is reported to be associated with more invasive treatment, advanced breast cancer, and economic deprivation, all of which are factors related to unemployment [24, 40, 41].
Regarding predictors of taking SL, our findings indicated that BCSs who had undergone surgery took SL more frequently than those who had undergone nonsurgical interventions. The distinction between BCSs who undergo surgery and those who do not suggests a relation to cancer stage (early or advanced) because almost all BCSs undergo surgery, except for those with stage IV cancer, in which distant metastasis is apparent. Previous studies have reported that breast cancer surgery is associated with SL lasting 1 month or longer [42, 43], and that the median duration of hospitalization among BCSs in Japan is about 6.79–10.37 days . Surgical treatment may result in challenging sequelae, including scar pain, fatigue, lymphedema, and reduced range of motion, particularly in the arm and chest region; these symptoms increase the time to RTW and are related to unemployment . Wennman-Larsen et al.  reported that arm morbidity shortly after surgery affected 10% of BCSs, and that 60% of these patients were on SL; SL was linked to arm morbidity, axillary clearance, and strenuous work posture. More invasive surgery is also related to more advanced breast cancer, which leads to more severe sequelae and longer SL . Petersson et al.  proposed that various side effects related to surgery impair work capacity and lead to longer SL in occupations requiring strenuous work postures.
This study did have some limitations. First, recall bias is possible given the nature of the self-report questionnaire design. In particular, as cognitive function may be adversely affected by some forms of treatment, some of the respondents may have been unable to remember when they had been diagnosed with breast cancer or to report how their work had changed after diagnosis. Second, this study was affected by survivorship bias, a form of selection bias, as BCSs who died before completing the questionnaire were excluded. Because BCSs who had been diagnosed with breast cancer within 1 year prior to participation in this study were excluded, we speculate that the resignation rate among BCSs was underestimated because of the death of patients who had left their jobs soon after diagnosis, especially in cases of advanced-stage disease. In addition, younger patients may have felt more comfortable than older patients given the online delivery and design of the survey. Third, SL is clearly stated as a right of employees according to company rules, and thus differs from merely being absent after using up annual paid leave. It might therefore be difficult to discuss the correlative factors of different situations all together. Fourth, the response rate was relatively low (10.4%) because a response was required within 2 days of receiving the questionnaire. It might be possible to increase the response rate by extending the response period. Finally, the sample size was small because a large number of respondents were ultimately excluded from analysis; further large-scale investigations are required to corroborate our results.
In the future, a prospective cohort study (such as an RTW intervention study) involving working BCSs in Japan is needed to clarify the association between symptoms and work-related factors among BCSs.