Occupational class and risk of renal cell cancer

Abstract Objectives We sought to examine the association between occupational class linked to job stress and the risk of renal cell cancer. To identify potential mediators, we additionally examined whether any observed associations persisted even after controlling for the contribution of stress‐related factors (eg, smoking, hypertension, and obesity). Methods Using nationwide inpatient records (1984 to 2016) from the Rosai Hospital group in Japan, we identified 3316 cases of renal cell cancer (excluding upper tract urothelial cancer) and 168 418 controls. We classified patients' occupational class (blue‐collar workers, service workers, professionals, and managers) and cross‐classified it by industry type (blue‐collar, service, and white‐collar) based on a standardized national classification. Unconditional logistic regression with multiple imputation was used for the analyses. Results A significantly elevated risk of renal cell cancer was found among men in higher occupational class (eg, professionals and managers). The elevated odds in male managers across all industries persisted even after controlling for smoking and alcohol consumption, with the association being more pronounced in blue‐collar industries (OR, 1.61; 95% CI, 1.34‐1.93). The association appeared to be mainly mediated by hypertension. Conclusion Occupational class is associated with the risk of renal cell cancer in men, particularly through modifiable risk factors.

association between hypertension and the risk of renal cell cancer has been previously undocumented in Japan.
Stress has long been hypothesized as a possible contributor to cancer risk via stress coping responses (ie, an increase in coping behaviors such as smoking or excess drinking), and/or direct physiological responses (eg, elevated blood pressure) that is partially mediated by activation of the sympathetic nervous system, inflammatory pathways, and the hypothalamic-pituitary-adrenal axis. 15,16 However, the empirical evidence linking various dimensions of stress to cancer incidence has remained inconsistent. 17,18 Regarding work-related stress, in the Nurses' Health Study, there was no association between multiple aspects of job stress, such as high demands and low control as well as low social support at work, and breast cancer or ovarian cancer. 19,20 Similarly, meta-analyses have not found an association between work stress and lung, colorectal, breast, or prostate cancer. 21 Yet no study to date has specifically investigated the relationship between stress because of work characteristics and renal cell cancer risk.
In Japanese society, higher occupational classes (managers and professionals) tend to report more job stress, 22,23 particularly following the collapse of the "economic bubble" in 1990. For example, Suzuki et al found that the occupational gradient in suicide in Japan reversed during the last 30 years. 22 Specifically, prior to the economic collapse of the asset bubble in 1991, suicide rates were higher among service, sales, and production workers. In the decades following the collapse, however, suicide rates have been higher among professional and managerial workers.
The distribution of job stress is markedly different in the Japanese workplace compared with the United States. For example, a recent study in Japan indicated that higher psychological distress in administrative and professional occupations is associated with increased cancer mortality at several sites. 24 Another study showed that the age-standardized suicide mortality rate increased among Japanese male administrative/managerial workers 22 between 1975 and 2005. In the same study, the lowest odds for suicide was observed among blue-collar production workers. 22 More recently, Tanaka et al 25 reported that the age-adjusted mortality rate for male managers increased across 12 types of occupation during the period of 1995 to 2010, which straddles the global economic crisis of 2008. While the magnitude of job stress across occupational classes is debated, 26,27 higher occupational class does indeed appear to be related to greater job stress in Japanese society, as indicated by the higher rates of suicide rates among managers and professionals in Japan. 22,23 Hence, in contrast to US/European studies, which typically show that job stress is higher among low-status occupations compared with high-status ones, the opposite pattern is found in Japan. Additionally, the prevalence of both hypertension and unipolar depression appeared to be higher in white-collar occupations compared with blue-collar occupations in Japan, 28,29 and hypertension appeared to be linked to job stress. 28 In the present study, we sought to examine the association between occupational class and renal cell cancer, assuming that occupational class is a proxy for work-related stress. 30,31 In addition, we assumed that occupational class is associated with stress-related factors (smoking, hypertension, and obesity), and that these may increase the risk for renal cell cancer. Therefore, we also tested whether any observed renal cell cancer risk associated with occupational class persisted even after controlling for the potential mediation by stressrelated factors.

| MATERIALS AND METHODS
We conducted a hospital-based case-control study using inpatient electronic medical records of the Rosai Hospital group run by the Japan Organization of Occupational Health and Safety, an FIGURE 1 Occupational class cross-classified with industrial cluster independent administrative agency. Details of the study database have been previously described. 32,33

| Cases and controls
The study subjects comprised 171 734 patients (3316 cases of renal cell cancer [excluding upper tract urothelial cancer] and 168 418 hospital controls) aged 20 years or older, admitted to hospitals between April 1984 and March 2016. According to available national statistics estimated with several high-quality local cancer registries in Japan, the total number of renal cell cancer cases in our data set represents 0.8% of the total incidence of kidney cancer (including upper tract urothelial cancer) in Japan for the years 1984 to 2013 (3033 of 357 993). 4 We excluded patients with the diagnosis of upper tract urothelial cancer or patients with preexisting cancer history from the cases. Controls were patients diagnosed with musculoskeletal diseases (ICD-9, 410-739 and ICD-10, M00-M99; 89%) and skin diseases (ICD-9, 680-709 and ICD-10, L00-L99; 11%). We assumed that these diagnoses selected for the control groups were not linked to work stress. 34

FIGURE 2
Odds ratios for renal cell cancer across different occupational classes stratified by sex. The odds ratio (dot) and 95% confidence interval (bar) were estimated by unconditional logistic regression with 5 imputed data. Male and female odds ratios were (A, C) adjusted for age and year of admission and (B, D) additionally adjusted for smoking and drinking. The numbers of cases and controls were, respectively, 2703 and 111 925 for men and 613 and 56 493 for women

| Occupational class defined by occupational and industrial category
The questionnaire included questions about the patients' current job and their 3 most recent ones (including age at starting and ending).

The occupations were coded with 3-digit codes in Japan Standard
Occupational Classification for occupation category and 3-digit codes in Japan Standard Industrial Classification for industry category. We selected the longest held job from the history for each patient.
Owing to the enormous variety of "longest held" jobs, we aggregated the occupations into 4 occupational classes, based on previous studies 26,27,35,36 : "blue-collar workers," "service and clerical workers," "professionals," and "managers." We also categorized the longest held occupations into 3 industrial clusters based on the methodology used in a previous study 37 : "blue-collar industry," "service and sales industry," and "white-collar industry" (Figure 1). We excluded those who were not actively engaged in paid employment (eg, homemakers, students, and unemployed) in the present study. In addition, we excluded female managers in the white-collar industry because we did not observe any renal cell cancer cases in that category.

| Covariates
Age and year of hospital admission were adjusted as confounding factors. To control potential changes in diagnosis and treatments over time, we adjusted for year of hospital admission. In mediation models, we included smoking and alcohol consumption, as well as potential stress-related factors such as hypertension, obesity, and diabetes, as mediators. We assumed that occupational class is associated with stress-related risk factors (smoking, hypertension, and obesity), and that these may increase the risk for renal cell cancer.

| Statistical analysis
Among study subjects, 11% did not provide information on occupational history, smoking, and alcohol consumption and 20% did not complete all data. The background characteristics differed between those with complete and incomplete data (Table S1), and excluding incomplete data may lead to biased inference. 38,39 To deal with missing data, we performed multiple imputation for missing data among the 171 734 study subjects using all data, including occupational class, smoking, and alcohol consumption. [38][39][40] Five imputed datasets were generated with multiple imputation by chained equations method 39,40 ; ; the following missing data were multiply imputed: occupational class (20 359, 12%), smoking (23 692, 14%), and alcohol consumption (48 608, 28%).
Using unconditional logistic regression with multiple imputation, we estimated the odds ratios (ORs) and 95% confidence intervals (CIs) for renal cell cancer in each occupational class, with blue-collar workers in the blue-collar industry as the reference group. We pooled the 5 ORs and 95% CIs obtained from each imputed dataset into one combined OR and 95% CI. 39

| RESULTS
Among men, those in higher occupational class (professionals and managers) had a significantly increased risk of renal cell cancer compared with blue-collar workers across all industry types ( Figure 2). In all 3 industries, men in the highest occupational groups, ie, managers, had significantly increased risk for renal cell cancer, with minimally adjusted OR ranging from 1.47 (for managers in the white-collar industry) to 1.62 (for managers in the blue-collar industry; Table 1). The observed increased OR for managers in all industries were not attenuated on adjustment for covariates and remained significantly associated with the risk for renal cell cancer on adjustment for covariates (adjusted OR ranged from 1.48 for managers in the white-collar industry to 1.61 for managers in the blue-collar industry, model 3; Table 1).
Among women, we observed marginal increases in the risks for managers ( Figure 2). The results in the minimal-adjusted and fulladjusted models were similar ( Table 1). The full-adjusted risk of managers and professionals in the service and sales industry were marginally elevated (model 3; Table 1).
In the subgroup analysis, the gradient of the ORs across occupational classes showed the same trend ( Figure 3). Among men, lifestyle-related diseases (hypertension, diabetes, and obesity) were independently associated with the risk for renal cell cancer (eg, hypertension, OR 1.36; 95% CI, 1.20−1.54; model 5; Table 2); the elevated risk for higher occupational class was attenuated largely by adjustment for hypertension (model 4). After fully adjusting for all potential mediating factors, the risk for higher occupational class was not significant (except for professionals in blue-collar and white-collar industries; model 5). Among women, the fully adjusted risk among higher occupational class workers was not significantly elevated ( Figure 3); however, the odds in the service and sales industries showed a trend suggesting  a positive occupational gradient pattern (ie, higher risk with higher occupational class; model 5; Table 2).
In sensitivity analyses, although the precise ORs and 95% CIs differed according to the analytic model and study population, the directions of the association (ie, higher risk with higher occupational class) were identical ( Figure 4 and Table S2). The result with complete data also showed the same pattern ( Figure S1). The correlation between hypertension, diabetes, and obesity were all significant (pairwise correlation; all P values < .001). The profile of patients treated in Rosai hospitals appeared to be nationally representative (Table S3). The average length of longest held jobs was over 20 years (Table S4).

| DISCUSSION
We found an elevated risk of renal cell cancer among high status occupations (managers and professionals) in men across all industry categories, suggesting that high job stress may partially be associated with the risk of renal cell cancer. We also found, for the first time, that hypertension is a relevant independent risk factor for renal cell cancer in Japan. Furthermore, the risk for renal cell cancer associated with higher occupational class was potentially mediated through the risk for renal cell cancer associated with stress-related risk factors-viz, hypertension as well as diabetes and obesity. A similar tendency was found for women working in the service and sales industry, although the effects were marginal.
Job stress may be related to risk of renal cell cancer through both direct and indirect causal pathways. The direct pathway posits that job stress increases risk through direct biological or mechanical stimulus to cancer stem cells (eg, oxidative stress). 41,42 Although the association between occupation and renal cell cancer was substantially explained by hypertension and other potential mediators (diabetes and obesity), some significant associations in blue-collar and white-collar industries persisted among men in the present study. This residual association suggests that the direct pathway may be partially pertinent for renal cell cancer.
The indirect pathway posits that job stress may increase the risk of renal cell cancer via risk factors potentially influenced by stressful occupations, eg, cigarette smoking or the prevalence of hypertension.
In fact, previous studies have suggested that psychological factors (eg, chronic or work environmental stress) can increase such lifestylerelated diseases. [43][44][45][46] In the present study, the prevalence of those who smoked more than 40 pack-years was higher in the managers than nonmanagers (25% versus 11%), and the prevalence of hypertension was greater in the managers (37% versus 27%).
In Japanese society, the concept of "hospitality" or omotenashi is emphasized in the service industry. Because of these expectations, those in managerial positions (or in the position of supervising other workers) may be particularly vulnerable to stress stemming from striving to meet customer expectations. In some instances, this situation has even led to death from overwork, referred to as karoshi. Such stress has been found to affect work-life balance among high occupational class workers. 47 By contrast, Whitehall studies showed that poorer health (eg, cardiovascular disease) is associated with low control at work, 48 which is usually the case for blue-collar workers in western contexts. Low control at work was also associated with less leisure-time physical activity. 49 Although our study is one of the largest case-control studies of renal cell cancer reported in Japan (3316 cases) and the profile of patients treated in Rosai hospitals appeared to be nationally representative 50 (Table S3), it represents less than 1% of the total incidence in the country as a whole. Hence, the generalizability of our findings to the rest of the country may be limited.
The strengths of our study include the large sample size and the detailed job information that enabled us to create occupational clas-  Table S4); therefore, in the sense of lifelong stress, our captured stress would be more relevant than stress measured at baseline only once in cohort studies. 21 In fact, a case-control study from Canada also found a significant association between job stress and cancer incidence at other sites. 52 Furthermore, a stressful working environment of the high occupational classes in Japan also enabled us to detect the association between higher occupational class, possibly linked to job stress, and the incidence of renal cell cancer. 22 There are some limitations in our study. First, in any hospitalbased case-control study, the selection of hospital controls may introduce selection bias in either direction (ie, toward or away from the null). However, sensitivity analysis, including controls diagnosed with all benign diseases (except malignant neoplasms) or only controls diagnosed with musculoskeletal disease, resulted in the same direction to increase the risk. Additionally, one-third of missing data may have introduced selection bias in either direction-even though the missing information were multiply imputed; however, the sensitivity analysis with complete data showed the same pattern. There might also be a potential recall bias in the self-reported information at the time of admission (eg, occupational history). However, the association of job stress and renal cell cancer was not widely known at that time. In addition, the questionnaires did not ask patients to report job stress, and the study subjects did not know the aim of our study. Therefore, the recall bias for occupational history may not be at play between the cases and controls, and this limitation might not affect our conclusion.
Second, occupational class is not a perfect proxy for job stress, and we could not directly assess job stress because our hospital electronic medical record data did not include an assessment of stress.
Higher occupational class may also reflect anxiety, depression, and other mental health conditions. 29 Kawakami et al also speculated that job commitment in these high positions might decrease the opportunities for investing in healthier behaviors such as leisure-time physical activity. 26 Physical activity has been found to be a protective factor for the risk of renal cell cancer. 53 A previous study found that the pattern of leisure-time physical activity differs in Japan compared with western contexts, viz, the highest levels of exercise were reported by clerical workers, while the lowest levels were reported among managerial workers and blue-collar workers. 54 In the same study, the highest levels of weekly physical activity, including occupational physical activity, were reported by blue-collar workers and the lowest levels among professional and managerial workers. 54

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.