2013 Volume 55 Issue 6 Pages 511-515
Objectives: The aim of this study was to assess the risk of developing bile duct cancer among workers in the other printing industry in comparison with workers in all industries in general. Methods: Prevalence of bile duct cancer was compared between workers in the printing industry and age-standardized controls in all other industries using the claims database of the Japan Health Insurance Association, which insures workers of small-medium sized employers of all industries. Results: Young (aged 30–49) male workers in the printing industry showed an elevated but insignificant standardized prevalence rate ratio (SPRR) for bile duct cancer in comparison with workers in all other industries (SPRR: 1.78; 95%CI: 0.63–5.00). The risk was higher for intrahepatic bile duct cancer but remained insignificant (SPRR: 3.03; 95%CI: 0.52–17.56). Conclusions: The sharply elevated risk of bile duct cancer observed among proof-printing workers of a printing factory in Osaka may not be generalizable to workers in the printing industry nationwide.
(J Occup Health 2013; 55: 511–515)
There have been growing interest and concern about the elevated risk of developing bile duct cancer among proof-printing industry workers since Kumagai reported about five cases of it in a printing factory in Osaka in May 20121). Subsequently, Kumagai surveyed 52 male proof-printing workers from the factory and identified 11 bile duct cancer patients, concluding that chemicals (1,2-dichloropropane, dichloromethane) were the most likely causes2). Also, in response to the 64 claims for workers compensation benefits made by the workers in the printing industry nationwide (as of February 28, 2013, of which 17 claims were made by workers of the factory in Osaka and 39 claims were by family members of workers who had died of bile duct cancer, with 7 claims overlapping3)), the Ministry of Health, Labour and Welfare (MHLW) organized a committee to investigate the causes of bile duct cancer by investigating cases in workers from the factory in Osaka. The committee surveyed 70 male proof-printing workers from the factory and identified 16 bile duct cancer patients concluding that 1,2-dichloropropane was the most likely cause4) (the discrepancy between Kumagai's report and the committee's report may be due to the different time windows: Kumagai surveyed workers who worked for at least one year between 1991 and 2006, but the committee surveyed workers between April 1991 and December 2012).
Kumagai encountered 11 cholangiocarcinoma (bile duct cancer) patients among 62 male workers employed at a printing company in Osaka. This unquestionably high SMR alerted many workers in the printing industry in general: a person would naturally be concerned about being subjected to a high risk of cancer. To address such concerns, it is necessary to compare the risk of cancer between the printing industry and other industries in general. The authors attempted to compare the prevalence of bile duct cancer between printing industry workers and workers in all other industries using claims data of the Japan Health Insurance Association (JHIA), which covers most employers in small-medium sized industries.
JHIA insures workers of small-medium sized employers (approximately 20 million beneficiaries) as well as their dependent family members (15 million). To the authors’ knowledge, the printing factory in Osaka is not insured by the JHIA (it is insured by a health insurance society), and hence workers and ex-workers of the factory are NOT counted in this study. Workers who quit their jobs may continue to enroll in the JHIA for a maximum of two years (voluntarily continuing beneficiaries). Enrollment is capped at the age of 74 because elderly individuals aged 75 years or older must enroll in a separate insurance system (the Health Care System for the Old-old).
The JHIA maintains a claims database and provides aggregate data as csv files for public use. The public use data contain the number of claims, number of days and amount of charges aggregated by calendar month, prefecture, status of beneficiaries (workers or family members), sex, ten-year age group, type of practice (inpatient, outpatient and dental) and the 119 classification of diagnoses5). Although it is not personally identifiable data, it is detailed enough for health economics research6).
The JHIA database contains 776,720,246 medical and dental claims (including inpatient and DPC claims) covering April 2009 thru March 2012 and includes all elements of health insurance claims data (diagnostic codes linkable to ICD10 coding, dates of diagnosis, distinction between definite and rule-out diagnoses, distinction between primary and secondary diagnoses, provider information and, detailed treatment information such as medication and clinical procedures). The database is personally identifiable and linkable to industrial classification of workplaces. The industrial classification consists of 42 categories including “printing and related industry”.
The database has been available since April 2009. We used the medical claims data for three years (April 2009-March 2012) because media coverage on the issue intensified beginning in May 2012, potentially biasing the utilization pattern of patients and the diagnosis patterns of doctors.
NumeratorThe number of unique patients with medical claims containing diagnoses C22.1 (intrahepatic bile duct cancer) and C24.0 (extrahepatic bile duct cancer) treated between April 2009 and March 2012 was used for numerator. Diagnoses include both primary and secondary diagnoses but excluded rule-out diagnoses. Exclusion of rule-out diagnoses is effective in reducing the false-positive rate particularly because the diagnostic category of cancer contains the highest percentage of rule-out diagnoses7). Age at diagnosis was determined by the date at initial diagnosis.
Cases include all beneficiaries enrolled in the JHIA at any time of the three year observation period. Beneficiaries who were once enrolled in the JHIA and were diagnosed as bile duct cancer after quitting the JHIA cannot be counted. This loss to the follow-up is a limitation of this study but such loss to the follow up will not bias the inter-industry comparison because such loss will occur in all industries equally.
DenominatorThe number of beneficiaries as of September 2010 was used for denominator. Each beneficiary was classified into 42 industrial categories. Voluntarily continuing beneficiaries were classified by the industrial classification of their previous workplaces.
AnalysisPrevalence was calculated using the above numerators and denominators. Age standardization was conducted using the beneficiaries in all other industries as a reference population. The expected prevalence was calculated for the printing industry by applying the sex- and age-specific prevalence of all other industries to the sex- and age-specific number of beneficiaries in the printing industry.
Ethics approvalThis study was approved by the Ethics Committee of Osaka City University.
Of approximately 35 million JHIA beneficiaries, there were a total of 201,937 workers and 168,420 dependent family members in the printing and related industry category as of September 2009, constituting approximately one percent of the JHIA's total enrollment. There were a total of 8,855 patients who were diagnosed as bile duct cancer at any time between April 2009 and March 2012, of whom 107 were in the printing and related industry category.
Expected number of patients for the printing and related industry category was calculated by applying the sex- and age-specific prevalence in all other industries to the sex- and age-specific number of beneficiaries in the printing and related industry category. Standardized prevalence rate ratio (SPRR) was calculated by applying the expected number of patients to the actually observed number of patients. The 95% confidence interval was calculated using Fisher's exact test (Table 1). Since concern was focused on the high incidence of bile duct cancer among young male workers, a separate table was created for the age group of 30–49 years old (Table 2).
There were five intrahepatic and five extrahepatic bile duct cancer patients observed among young male workers in the printing and related industry category, representing 3.03 times (95%CI: 0.52–17.56) and 1.26 times (95%CI: 0.34–4.71) more than the expected number of patients. Overall, young male workers in the printing industry showed an elevated but insignificant SPRR for bile duct cancer in comparison with all other industries (SPRR, 1.78; 95%CI: 0.63–5.00). However, none of them reached statistical significance due to the small sample size.
| Workers | Family members | Total | |||||
|---|---|---|---|---|---|---|---|
| M | F | MF | M | F | MF | ||
| Observed number of patients | |||||||
| C22 (intrahepatic) | 24 | 3 | 27 | 2 | 5 | 7 | 34 |
| C24 (extrahepatic) | 42 | 7 | 49 | 7 | 17 | 24 | 73 |
| Total | 66 | 10 | 76 | 9 | 22 | 31 | 107 |
| Expected number of patients | |||||||
| C22 (intrahepatic) | 13.40 | 2.80 | 15.91 | 1.24 | 5.71 | 7.14 | 23.02 |
| C24 (extrahepatic) | 37.08 | 7.14 | 43.62 | 4.29 | 14.48 | 19.31 | 62.96 |
| Total | 50.48 | 9.94 | 59.53 | 5.53 | 20.18 | 26.45 | 85.99 |
| Standardized prevalence rate ratio (observed/expected) | |||||||
| Upper limit of 95% CI | 3.49 | 5.45 | 3.15 | 15.19 | 2.91 | 2.78 | 2.51 |
| C22 (intrahepatic) | 1.79 | 1.07 | 1.70 | 1.62 | 0.88 | 0.98 | 1.48 |
| Lower limit of 95% CI | 0.92 | 0.21 | 0.91 | 0.17 | 0.26 | 0.35 | 0.87 |
| Upper limit of 95% CI | 1.76 | 2.78 | 1.69 | 5.43 | 2.37 | 2.26 | 1.62 |
| C24 (extrahepatic) | 1.13 | 0.98 | 1.12 | 1.63 | 1.17 | 1.24 | 1.16 |
| Lower limit of 95% CI | 0.73 | 0.35 | 0.75 | 0.49 | 0.58 | 0.68 | 0.83 |
| Upper limit of 95% CI | 1.89 | 2.42 | 1.79 | 4.70 | 1.99 | 1.97 | 1.65 |
| Total | 1.31 | 1.01 | 1.28 | 1.63 | 1.09 | 1.17 | 1.24 |
| Lower limit of 95% CI | 0.91 | 0.42 | 0.91 | 0.56 | 0.60 | 0.70 | 0.94 |
| Workers | Family members | Total | |||||
|---|---|---|---|---|---|---|---|
| M | F | MF | M | F | MF | ||
| Observed number of patients | |||||||
| C22 (intrahepatic) | 5 | 1 | 6 | 6 | |||
| C24 (extrahepatic) | 5 | 1 | 6 | 1 | 2 | 3 | 9 |
| Total | 10 | 2 | 12 | 1 | 2 | 3 | 15 |
| Expected number of patients | |||||||
| C22 (intrahepatic) | 1.65 | 0.37 | 1.92 | 0.06 | 0.52 | 0.63 | 2.55 |
| C24 (extrahepatic) | 3.97 | 0.94 | 4.68 | 0.11 | 1.13 | 1.30 | 6.02 |
| Total | 5.62 | 1.31 | 6.60 | 0.17 | 1.65 | 1.93 | 8.56 |
| Standardized prevalence rate ratio (observed/expected) | |||||||
| Upper limit of 95% CI | 17.56 | 115.79 | 15.87 | 10.20 | |||
| C22 (intrahepatic) | 3.03 | 2.69 | 3.12 | 2.35 | |||
| Lower limit of 95% CI | 0.52 | 0.06 | 0.62 | 0.54 | |||
| Upper limit of 95% CI | 4.71 | 17.79 | 4.30 | 4,560.87 | 17.83 | 18.00 | 4.20 |
| C24 (extrahepatic) | 1.26 | 1.06 | 1.28 | 9.07 | 1.77 | 2.30 | 1.50 |
| Lower limit of 95% CI | 0.34 | 0.06 | 0.38 | 0.02 | 0.18 | 0.29 | 0.53 |
| Upper limit of 95% CI | 5.00 | 13.79 | 4.70 | 932.89 | 9.50 | 9.50 | 4.05 |
| Total | 1.78 | 1.52 | 1.82 | 5.74 | 1.21 | 1.56 | 1.75 |
| Lower limit of 95% CI | 0.63 | 0.17 | 0.70 | 0.04 | 0.15 | 0.25 | 0.76 |
The SPRR for both sexes combined among young workers in Table 2 showed a larger ratio than the ratio for each sex (the SPRR of bile duct for both sexes was 1.82, while that for males was 1.78 and that for females was 1.52). This seemingly odd phenomenon is due to the sex imbalance in the number of workers in the printing industry. When the age-specific prevalence for all other industries was applied to the printing industry, which showed a disproportionately higher SPRR for male workers, the sex imbalance caused the expected number of bile duct cancer in the printing industry to be smaller than the sum of the expected number of each sex (the expected number of both sexes: 6.60, smaller than the sum of male: 5.62 and female: 1.31).
There has been intensive public interest and concern regarding the suspected risk of bile duct cancer among printing industry workers since the first case-series were reported with regard to a printing factory in Osaka in May 2012. Although the report was about proof-printing workers at a certain factory, the public was alerted that the same phenomena might be happening among workers in the same industry nationwide. The MHLW quickly conducted a questionnaire survey among a total of 18,131 printing factories nationwide and announced that they received reports of 22 cases of bile duct cancer (including 12 deaths) from 14,267 factories (response rate: 78.7%)8). However, it is difficult ascertain the causality because 20 of the cases were in individuals over 50 years old.
Internationally, there have been sporadic reports concerning the occupational risks of the printing industry. A British researcher, prompted by an anecdotal report of a cluster of cases of bladder cancer in a newspaper factory in Manchester, conducted a cohort study among workers of a printing factory. Although the results did not support the occupational risk of bladder cancer, they did demonstrate elevated all-cause mortality among them9). Danish researchers following a cohort of printing workers demonstrated elevated risks of lung, bladder, renal pelvis and primary liver cancers among printing workers10). French researchers also demonstrated elevated risks of lung and esophageal cancer among workers of an offset printing plant11).
However, none of the previous reports from around the world demonstrated an elevated risk of bile duct cancer among printing workers in particular. So far, the evidence has been confined to a case-series from a single factory in Osaka. To demonstrate if any elevated risk of bile duct cancer exists in the printing industry in Japan in general, it is necessary to compare age-standardized prevalence of the disease between the target industry and other industries, which is difficult to achieve because one has to cover workers in all industries. A large-scale administrative database would provide unbiased and reliable estimates that could be used to perform these comparisons. The JHIA is a single large health insurer insuring workers of small-medium sized employers of all industries and therefore appears to be most appropriate data source for comparison of the prevalence of specific diseases among different industries.
Since the JHIA database is an administrative database not intended for epidemiological studies, it has limitations. First, diagnoses contained in health insurance claims are not definitive diagnoses confirmed by doctors. The authors avoided overdiagnosis by excluding diagnoses with rule-out modifiers, but concerns about the validity of diagnoses remain. Second, not all workers working at workplaces classified into the printing and related industry category are exposed to hazardous environments. They include clerical workers of a printing company thereby diluting the effects of occupational exposures.
The results showed a slightly elevated prevalence of intrahepatic bile duct cancer among male workers in the printing industry (SPRR, 1.79; 95%CI: 0.92–3.49). If limited to younger age (30–49), the SPRR was higher but remained insignificant (SPRR, 3.03; 95%CI: 0.62–17.56).
Our results demonstrated that the elevated risk of bile duct cancer observed in the proof-printing workers of the Osaka factory may not be generalizable to all workers in the printing industry. The relation between work and bile duct cancer should be evaluated in the future by estimation of occupational exposure of causative chemicals on an individual basis.
Acknowledgment: This study was supported by the Ministry of Health, Labor and Welfare research grant for “Epidemiological study on causes of bile duct cancer observed among workers in the printing industry (PI: Ginji Endo)”.
