Document 2JggDxbba3BN8m0m725wDKNkL
Occupational History and Exposure and the Risk of Adult Leukemia in Shanghai
OLUFEMI J. ADEGOKE, MBCHB, DRPH, AARON BLAIR, PHD, XIAO OU SHU, MD, PHD1, MAUREEN SANDERSON, PHD2, FAN JIN, MD, MUSTAFA DOSEMECI, PHD, CHERYL L. ADDY, PHD, AND WEI ZHENG, MD, PHD1
PURPOSE: Evaluation of the association of selected occupational exposures with leukemia risk. METHODS: Population-based case-control study of 486 leukemia subjects and 502 healthy controls residing in Shanghai from 1987 to 1989. Adjusted odds ratios (OR) were calculated for the association between occupational factors and leukemia risk. RESULTS: Significant increase in leukemia risk was observed in chemical manufacturing industry workers (OR 3.1, 95% CI 1.09.8). Increased risks for leukemia were observed from self-reported exposures to benzene (OR 1.7, 95% CI 1.12.6), radioactive materials (OR 3.7, 95% CI 1.3 10.2), synthetic fiber dust (OR 2.0, 95% CI 1.23.5), and toluene (OR 1.6, 95% CI 1.02.5). Doseresponse relations of leukemia risk was observed with the duration of exposure to benzene (OR 3.3, 95% CI 1.66.9 for 15 years exposure; p for trend 0.01), radioactive materials (OR 5.2, 95% CI 1.124.7 for 15 years exposure; p for trend 0.02), paints (OR 2.3, 95% CI 1.24.7 for 15 years exposure; p for trend 0.09), and toluene (OR 2.9, 95% CI 1.36.7 for 15 years exposure; p for trend 0.02). CONCLUSIONS: Adult leukemia risk may be associated with working in the chemical industry, and exposure to benzene, synthetic fiber dust, radioactive materials, and toluene in the study population. Ann Epidemiol 2003;13:485494. 2003 Elsevier Inc. All rights reserved.
KEY WORDS: Leukemia, Occupational Exposures, Risk Factor, Case-control Study.
.
INTRODUCTION
Though a number of occupational epidemiologic studies of leukemia have been conducted, there are few established occupational risk factors other than benzene and radiation exposure (1, 2). Most previous studies were small, occupational exposures were not systematically assessed, and were unable to control for potential confounding factors (36). Some studies evaluated only a single exposure (e.g. benzene,
Portions of this work constituted partial fulfillment for a doctoral degree in epidemiology and biostatistics at the University of South Carolina, Columbia, SC by O.J.A.
From the Department of Epidemiology and Biostatistics, Norman J. Arnold School of Public Health, University of South Carolina, Columbia, SC 29208 (O.J.A., X.O.S., M.S, C.L.A., W.Z.); Occupational Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, MD 208509 (A.B., M.D.); and Department of Epidemiology, Shanghai Cancer Institute, Shanghai 200032, People's Republic of China (F.J.).
1 Present address: Dept. of Medicine and Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, TN, 37232.
2 Present address: University of Texas School of Public Health at Brownsville, Brownsville, TX 78520.
Address correspondence to: Olufemi J. Adegoke, MBChB, DrPH, MPH, Department of Surgery, School of Medicine, Meharry Medical College, 1005 Dr. D.B. Todd Jr. Blvd., Nashville, TN 37208. Tel.: (615) 327-5689; Fax: (615) 327-5579. E-mail: oadegoke@mmc.edu
Received February 20, 2002; accepted December 16, 2002.
2003 Elsevier Inc. All rights reserved. 360 Park Avenue South, New York, NY 10010
organic solvents, radiation) (68). Moreover, many previous studies did not evaluate exposures by type of leukemia (9). Finally, few studies have been conducted in China.
Shanghai is the largest industrialized city on the east coast of China. The incidence rates for leukemia in Shanghai ranged from 3.5 to 5.0 per 100,000 person-years between 1988 and 1992, which were similar to overall rates in China (10, 11). Since few studies have been conducted in China, a case-control study of leukemia was conducted in Shanghai to evaluate familial, environmental/ occupational, medical, and therapy-related risk factors. The industrial diversity in Shanghai provides a unique opportunity to evaluate a number of suspected occupational risk factors for leukemia.
METHODS
Study Participants The methods for this study have been described in detail in an earlier publication (12). Cases were residents of urban Shanghai diagnosed with leukemia at age 15 or older (ICD9 codes 204208 [WHO, 1997]) who were reported to the Shanghai Cancer Registry from June 1, 1987 through
1047-2797/03/$see front matter doi:10.1016/S1047-2797(03)00037-1
GARABRANT/BERRYMAN 00054
486 Adegoke et al.
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
AEP Vol. 13, No. 7 August 2003: 485494
Selected Abbreviations and Acronyms
CI confidence interval OR odds ratio ICD-9 International Classification of Diseases-9 WHO World Health Organization SAS Statistical Analysis Software ALL aute lymphocytic leukemia ANLL acute non-lymphocytic leukemia CML chronic myeloid leukemia CLL chronic lymphocytic leukemia EMF electromagnetic field JEM job-exposure matrix
August 31, 1989. A total of 532 eligible cases were identified. Interviews were completed with 486 (91.4% of the eligible cases). Interviews were obtained directly with 254 cases, while surrogates provided information for 194 cases that were deceased and 38 cases too ill to be interviewed. Forty-eight percent of the surrogates were spouses and 50% other first-degree relatives. Of the eligible cases, 35 (6.5%) could not be located and 11 (2.1%) refused to participate.
Controls were randomly selected from the general urban Shanghai population using the comprehensive, population-based Resident Registry. The Resident Registry maintains a personal registry card for each adult resident (over age 15) in urban Shanghai. The information on each resident includes name, address, date of birth, sex, and other demographic factors. The likely age (5-year intervals) and sex distribution of cases was determined using leukemia incidence data from 1984 to 1985, which was the most recently available information from the Shanghai Cancer Registry at the start of the study. Based on this distribution, 502 potential controls were randomly selected from residents as listed in the Resident Registry. If the first control selected was ineligible or refused to be interviewed, a second control was then chosen using the same method. Only 30 replacement controls (6.0%) were required. These were used because the first controls could not be located (12 out of 30), died prior to being interviewed (8 out of 30), or for other reasons (10 out of 30). The response rate among controls was 94 percent.
Exposures
Trained interviewers used standardized Chinese questionnaires to obtain information on demographic factors, occupational exposures, lifestyle, occupational history, family history, dietary habits, tobacco smoking, education, diagnostic X-ray procedures, specific medications and medically related topics. Each subject was asked to provide the information for each job he/she held for at least 3 years during his/her lifetime. Self-reported information was also sought on exposure to benzene, gasoline, pesticides, synthetic fiber dust, toluene, radioactive materials (exposure to X-rays and
diagnostic radiation), electromagnetic fields (EMF), asbestos, paints, and other organic solvents. Excessive risks for leukemia have been associated with some of these exposures by others (1316). Subjects who reported having these exposures were questioned further for information on the frequency and duration of exposure. The job titles were classified using industrial and occupational headings defined for the 1982 Chinese population census (17). The census code grouped industries and occupations according to three levels of increasing detail. The major headings reflected general industry or occupation groupings, whereas the second and third levels provided finer, more specific classifications of each industry or occupation (17).
A job-exposure matrix (JEM), along with lifetime job history, was used to indicate the probability of exposure to benzene, other organic solvents, pesticides, and EMF. The JEM was developed by Mustafa Dosemeci at the US National Cancer Institute for occupations commonly held by Chinese in Shanghai. It assigned an overall probability of exposure of none (0), low (1), medium (2), or high (3) to each job using an algorithm based on a combination of the three-digit job title and three-digit industry codes.
Odds ratios for specific jobs were calculated by comparing "ever" vs. "never" holding a specific job for at least 3 years during his/her lifetime. Only job titles with at least 15 cases and controls combined were analyzed. One-, two-, and three-digit job title codes were analyzed, but numbers for three-digit job codes were too small for presentation. For selfreport of exposures, analyses were by "ever exposed" vs. "never exposed" and by the total number of years spent on jobs with exposure, i.e. 15 years, and 15 years. Finally, the JEM-based exposure assessment was combined with the self-reported exposures and categorized into "ever" vs. "never". For each exposure, the "never" group was those with "no" job probability of exposure on the JEM and who also reported "no" on self-report. The "ever" group was made up of any one with "yes" on self-report or who had a "low", "medium", or "high" probability of job-exposure on the JEM.
We evaluated the relationship between occupational exposures in jobs held 3 years or more and total leukemia, as well as with subtypes (acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), chronic lymphoblastic leukemia (CLL), and chronic myeloid leukemia (CML)). Odds ratios (OR) are used in our analyses to approximate relative risk (RR) because of the rarity of leukemia. ORs and 95% confidence intervals (CI) were calculated using multivariate logistic regression (18) for all types of leukemia combined. Analyses by subtypes of leukemia were conducted using polychotomous logistic regression where the subtypes of leukemia are treated as discrete variables as there was more than one case group compared with the same control referent group. The associations between leukemia and occupational exposures were evaluated by gender and interview
GARABRANT/BERRYMAN 00055
AEP Vol. 13, No. 7 August 2003: 485494
Adegoke et al. 487
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
type, i.e., direct vs. surrogate. All odds ratios were adjusted for age, gender and income. Age was entered as a continuous variable and income was categorized into tertiles. All analyses done were performed using Statistical Analysis Software (SAS).
RESULTS
Cases and controls were well matched for age and gender. The mean age for cases was 51.1 years and 50.8 years for controls. The leukemia cases were similar to controls in years of education, but had a slightly lower household income (Table 1). There was some variation in demographic factors among the different leukemia subtypes. The maleto-female ratio ranged from 1.0 for AML to 1.6 for CLL. The ALL cases tended to be younger than AML, CLL, and CML participants. The CLL participants were mostly over age 65. Because of the small number of the CLL subtype group (N 21), findings on CLL from the analysis are not shown in the tables, but are summarized briefly in the text.
The odds ratios for the risk of leukemia by lifetime occupations are shown in Table 2. The risk for leukemia was increased significantly among those who worked in the chemical manufacturing industry (OR 3.1, 95% CI 1.09.8). Borderline significant excesses for leukemia were seen in government officials, leaders and managers (OR 1.7, 95% CI 0.93.2), and textile workers (OR 1.4, 95% CI 0.92.3). Non-significant increases in risk for leukemia were also seen among medical and public health professionals (OR 1.4, 95% CI 0.63.2),
primary metal workers (OR 1.5, 95% CI 0.63.7), and construction workers (OR 1.8, 95% CI 0.74.5). The risk for all leukemias was significantly reduced among participants who worked as metal machine operators (OR 0.4, 95% CI 0.20.9). The small numbers for all these job titles, however, prevented analyses by subtypes of leukemia.
The odds ratios for the associations between ever exposed to selected self-reported occupational exposures, duration of exposure, and the risk of all leukemias and leukemia subtypes were obtained by polychotomous logistic regression and are presented in Table 3. The OR among those ever exposed to benzene was 1.7 (95% CI 1.12.6) for total leukemia. The risk was increased with the duration of exposure to benzene (p for trend 0.01) to an OR of 3.3 (95% CI 1.66.9) among those with 15 years or over of exposure. The ever exposed category had a significantly increased risk for CML (OR 2.5, 95% CI 1.34.9) and a nonsignificant increase for ALL (OR 1.5, 95% CI 0.7 3.3) and AML (OR 1.4, 95% CI 0.82.3). However, elevated risk associated with long-term benzene exposure (15 years) was observed for all leukemia subtypes.
Among those ever exposed to radioactive materials, the ORs for all leukemias combined, ALL and CML were statistically significant, but not significant for AML. Long-term exposure of 15 or more years to radioactive materials was significantly associated with an increased risk for all leukemias. The duration of exposure to radioactive materials increased the risk to all leukemias (p for trend 0.02). Significant increases in risk for ALL and CML were also noted among subjects with long-term exposure to radioactive materials. However, numbers were small for some duration and leukemia subtypes categories.
TABLE 1. Percentage distribution of controls and adult leukemia participants by subtypes according to selected demographic factors
Leukemia cases
Characteristics
Sex Male Female
Age group (yrs) 35 3554 5564 65
Education (yrs) 6 611 12
Income tertile T1 (low) T2 T3
Controls (n 502)
52.8 47.2
25.1 24.7 23.1 27.1
31.7 46.0 22.3
34.3 36.1 29.7
Total (n 486)
52.7 47.3
24.3 23.9 24.9 27.0
31.3 44.5 24.2
47.1 33.1 19.8
ALL (n 81)
54.3 45.7
34.6 27.2 22.2 16.1
23.8 55.0 21.3
46.9 39.5 13.6
CLL (n 21)
61.9 38.1
9.5 4.8 9.5 76.2
47.6 33.3 19.1
57.1 19.1 23.8
AML (n 236)
48.7 51.3
24.2 23.3 25.4 27.1
30.5 44.9 24.6
50.0 30.5 19.5
CML (n 79)
58.2 41.8
20.3 34.2 20.3 25.3
29.5 42.3 28.2
41.8 39.2 19.0
GARABRANT/BERRYMAN 00056
488 Adegoke et al.
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
AEP Vol. 13, No. 7 August 2003: 485494
TABLE 2. Odds ratios of leukemia by lifetime occupation
Codea
Lifetime occupationb
No. of exposed case/controlc
ORd
CIe
I.
01 02 03 04 II.
06 07 III. 08 09 IV. 14 15 V.
16 VII.
18 19 20 21 22 28 29 30 31 32 34 35 VIII. 38 IX. 40 41 X.
Professional and technical workers, scientists, and research workers
Scientific research workers Medical and Public Health Financial and related workers Teachers Leaders of government agencies, branches,
parties, mass organizations and Businesses Officials, leaders and managers Clerical and related workers Commercial workers (sales workers) Salesman Buyers Service workers Other service workers Cooks Agricultural workers, foresters, animal
husbandry workers, fishermen, and hunters Farmers Production, transportation, and related
workers (manufacturing) Primary metal workers Chemical processors/related workers Rubber and plastics product makers Textile workers Apparel Tool makers Metal machine operators Machine assembler Electrician Welders and sheet metal Production inspector Package and warehouse Construction Construction Utility and transport Water and rail transportation Auto and truck drivers Housewives
aOccupational codes based on 1982 Chinese population census (19). bLifetime job. cBased on ten or more total cases and controls employed in the occupation. dAll OR (odds ratio) adjusted for age, sex and income. e95% confidence interval.
79/76
22/19 15/11 17/21 20/22 46/44
27/21 19/23 30/39 21/32 9/7 32/43 15/19 9/13 12/15
12/15 194/193
11/8 13/4 10/8 42/33 14/12 15/11 10/23 17/21 10/13 9/7 9/11 16/16 27/17 13/7 27/41 11/23 8/10 22/20
1.2 0.81.7
1.2 0.62.3 1.4 0.63.2 0.9 0.51.8 1.1 0.62.1 1.3 0.82.0
1.7 0.93.2 0.9 0.51.7 0.8 0.51.4 0.7 0.41.3 1.4 0.53.9 0.7 0.41.1 0.7 0.31.4 0.7 0.31.6 0.8 0.41.7
0.8 0.41.7 1.1 0.81.4
1.5 0.63.7 3.1 1.09.8 1.3 0.53.3 1.4 0.92.3 1.2 0.52.7 1.3 0.63.0 0.4 0.20.9 0.8 0.41.6 0.7 0.31.7 1.3 0.53.6 0.9 0.42.2 1.1 0.52.3 1.5 0.82.9 1.8 0.74.5 0.7 0.41.1 0.5 0.21.1 0.8 0.32.0 1.0 0.51.9
Those ever exposed to toluene had a significant 60% elevation in risk for all leukemias, but non-significant excesses for ALL, AML, and CML. Duration of exposure to toluene was also associated with increasing risk for leukemia (p for trend 0.02) with 15 or more years of exposure.
A borderline significant excess was observed among ever exposed compared with never exposed to paints for CML. Long-term exposure to paints was associated with significant elevations for all leukemias, ALL, and CML, while a nonsignificant elevation was observed for AML. The trend for duration of exposure to paints and association with leukemia and the subtypes were not significant.
Exposure to other organic solvents increased the risk for CML, but not for ALL or AML. Long-term exposure to organic solvents was associated with non-significant risks for ALL and CML. Trends by duration of exposure were generally unimpressive for organic solvents exposure and other exposures.
Information on duration of exposure was not available for exposure to synthetic fiber dust, or asbestos. Significant excesses for all leukemias, AML, and ALL, but borderline for CML, were observed among subjects ever exposed to synthetic fiber dust. On further analyses, we observed that the association of synthetic fiber dust with an increased risk
GARABRANT/BERRYMAN 00057
AEP Vol. 13, No. 7 August 2003: 485494
Adegoke et al. 489
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
TABLE 3. Odds ratios of leukemia by selected occupational exposures on lifetime job from self-report among all participants
Exposuresa
Controls (n 502)
All leukemias (n 486)
Cases ORb CIc
ALL (n 81)
Cases ORb
CIc
AML (n 236)
Cases ORb
CIc
CML (n 79)
Cases ORb
CIc
Benzene
Never
462
Ever 40
15 yrs
30
15 yrs
10
Trend test
Other organic solvents
Never
413
Ever 89
15 yrs
57
15 yrs
32
Trend test
EMF
Never
358
Ever 144
15 yrs
93
15 yrs
51
Trend test Pesticidesa
Never
479
Ever 23
10 yrs
18
10 yrs
5
Trend test
Gasoline
Never
379
Ever 123
15 yrs
68
15 yrs
55
Trend test
Radioactive materials
Never
497
Ever 5
15 yrs
3
15 yrs
2
Trend test
Paints
Never
452
Ever 50
15 yrs
38
15 yrs
12
Trend test
Toluene
Never
468
Ever 34
15 yrs
26
15 yrs
8
Trend test Synthetic fiber dustd
Never
479
Ever Asbestosd
23
Never
491
Ever 11
424 62 34 28 p 0.01
1.0 -- 1.7 1.12.6 1.2 0.72.1 3.3 1.66.9
72 9 4 5 p 0.12
1.0 -- 1.5 0.73.3 0.8 0.32.4 3.9 1.311.8
212 24 12 12
p 0.01
1.0 -- 1.4 0.82.3 0.8 0.41.7 2.9 1.27.0
65 14 7 7 p 0.01
1.0 -- 2.5 1.34.9 1.7 0.74.0 5.0 1.813.9
396 90 54 36 p 0.62
1.0 -- 1.1 0.81.5 1.0 0.71.5 1.2 0.82.1
69 12 3 9 p 0.82
1.0 -- 0.7 0.41.4 0.2 0.10.8 2.0 0.94.4
192 44 29 15
p 0.86
1.0 -- 1.1 0.81.7 1.1 0.71.9 1.1 0.62.1
58 21 13 7 p 0.08
1.0 -- 1.7 1.03.0 1.7 0.83.4 1.8 0.84.1
361 1.0 -- 125 0.9 0.71.2 64 0.7 0.51.0 61 1.2 0.81.8 p 0.91
55 26 13 13 p 0.28
1.0 -- 1.1 0.71.8 0.8 0.41.5 1.8 0.93.6
182 1.0 -- 54 0.8 0.51.1 31 0.7 0.41.1 23 0.9 0.61.6
p 0.23
49 1.0 -- 30 1.6 0.92.6 13 1.0 0.52.0 17 2.4 1.34.5 p 0.02
464 1.0 -- 22 0.9 0.51.7 15 0.8 0.41.7 7 1.4 0.44.3 p 0.81
79 2 1 1 p 0.55
1.0 -- 0.4 0.11.8 0.3 0.11.9 0.9 0.18.4
225 1.0 -- 11 1.0 0.52.0 7 0.8 0.31.9 4 1.6 0.46.1
p 0.73
74 1.0 -- 5 1.3 0.53.7 13 1.0 0.52.0 1 1.2 0.110.5 p 0.56
379 1.0 -- 107 0.9 0.61.2 55 0.8 0.51.2 53 1.0 0.61.4 p 0.58
66 15 9 6 p 0.20
1.0 -- 0.6 0.31.1 0.6 0.31.2 0.6 0.31.5
187 1.0 -- 49 0.8 0.51.2 28 0.8 0.51.3 22 0.8 0.51.4
p 0.33
56 1.0 -- 23 1.2 0.72.1 10 0.9 0.42.0 13 1.5 0.83.0 p 0.25
471 1.0 -- 15 3.7 1.310.2 7 2.7 0.710.5 8 5.2 1.124.7 p 0.02
78 1.0 -- 3 4.4 1.019.2 0 0.0 -- 3 14.1 2.289.9 p 0.01
230 1.0 -- 6 3.2 0.910.5 3 2.5 0.512.5 3 4.2 0.725.8
p 0.10
75 1.0 -- 4 5.7 1.522.0 2 4.3 0.726.5 2 7.9 1.158.5 p 0.01
429 1.0 -- 57 1.2 0.81.7 30 0.8 0.51.3 27 2.3 1.24.7 p 0.09
71 10 4 6 p 0.15
1.0 -- 1.1 0.52.4 0.5 0.21.6 3.4 1.29.5
216 1.0 -- 20 0.8 0.51.4 10 0.5 0.21.1 10 1.8 0.74.2
p 0.92
66 1.0 -- 13 1.7 0.93.2 6 1.0 0.42.5 7 3.7 1.49.9 p 0.02
437 1.0 -- 49 1.6 1.02.5 28 1.1 0.72.0 21 2.9 1.36.7 p 0.02
74 7 3 4 p 0.22
1.0 -- 1.3 0.63.1 0.6 0.22.2 3.7 1.112.9
215 1.0 -- 21 1.4 0.82.4 12 1.0 0.52.0 9 2.6 1.06.9
p 0.14
69 1.0 -- 10 2.0 0.94.2 6 1.5 0.63.9 4 3.4 1.011.7 p 0.03
444 1.0 -- 42 2.0 1.23.5
73 1.0 -- 8 2.5 1.05.9
213 1.0 -- 23 2.2 1.24.1
72 1.0 -- 7 2.3 0.95.6
475 1.0 -- 11 1.1 0.52.6
78 1.0 -- 3 1.6 0.46.0
231 1.0 -- 5 1.1 0.43.2
76 1.0 -- 3 1.7 0.56.2
aDuration of exposure categories different because of small numbers. bAll OR (odds ratio) adjusted for age, sex, and income. c95% confidence interval. dNo duration of exposure information collected for synthetic fiber use and asbestos.
GARABRANT/BERRYMAN 00058
490 Adegoke et al.
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
AEP Vol. 13, No. 7 August 2003: 485494
for all leukemia was higher among non-textile workers compared with textile workers. Exposure to asbestos showed non-significant increases in risk for ALL and CML. A few other non-significant excesses were also observed for CML from exposure to EMF, pesticides, and gasoline.
When the analysis was restricted to directly interviewed subjects only (data not shown in tables) the ORs typically increased slightly compared with those that also included information from surrogate respondents. For example, the ORs for all leukemia for "ever" exposed to benzene increased from 1.7 to 2.3 (both significant) and for ALL increased from 1.5 to 2.0 (both non-significant). Since direct ORs were in general similar to the overall ORs, we have chosen to present the overall ORs (i.e., both direct and proxy interviews).
The results of gender-specific analyses for selected occupational exposures are shown in Table 4. Among the men, statistically significantly elevated risks occurred for leukemia among those ever exposed compared with those never exposed to radioactive materials (OR 7.5, 95% CI = 1.6 34.1) and synthetic fiber dust (OR 3.2, 95% CI 1.0
10.2), and the OR associated with toluene OR 1.6, 95% CI 0.91.9) was of borderline significance. Risk for leukemia among men reporting exposure to benzene was elevated, not significantly. For women, significantly elevated risks were observed for benzene exposure (OR = 2.3, 95% CI 1.24.6) and synthetic fiber dust exposure (OR 1.8, 95% CI 1.03.3) and there was a borderline significant risk for paints. Risks among women reporting exposure to pesticides and toluene were elevated, but not significant. We have not reported the results of gender-specific analyses for leukemia subtypes due to small numbers.
To refine the exposure classification, we incorporated information from both the JEM and self-reports in defining exposure status. The results are presented in Table 5. Individuals in the unexposed category in both exposure systems, i.e., the No/No group, served as the reference. Ever exposed included those classified as exposed by either the self-report or the JEM. The results from this analysis were very similar to those presented in Table 3 based on self-reported exposure information alone.
TABLE 4. Gender-specific odds ratios of leukemia by selected occupational exposures on lifetime job from self-report for all leukemias among all participants
Exposures
Men (n 521)
Women (n 467)
Controls
Cases
ORa
CIb
Controls
Cases
ORa
Benzene Never Ever
Other organic solvents Never Ever
EMF Never Ever
Pesticides Never Ever
Gasoline Never Ever
Radioactive materials Never Ever
Synthetic fiber dust Never Ever
Asbestos Never Ever
Paints Never Ever
Toluene Never Ever
238 27
214 51
179 86
248 17
178 87
263 2
261 4
257 8
230 35
244 21
aAll OR (odds ratio) adjusted for age and income. b95% confidence interval.
222 34
207 49
186 70
246 10
187 69
244 12
245 11
248 8
224 32
227 29
1.0 -- 1.5 0.82.5
1.0 -- 1.0 0.71.6
1.0 -- 0.8 0.51.2
1.0 -- 0.6 0.31.3
1.0 -- 0.7 0.51.1
1.0 -- 7.5 1.634.1
1.0 -- 3.2 1.010.2
1.0 -- 1.1 0.42.9
1.0 -- 0.9 0.51.5
1.0 -- 1.6 0.92.9
224 13
199 38
179 58
231 6
201 36
234 3
218 19
234 3
205 15
224 13
202 1.0 28 2.3
189 1.0 41 1.2
175 1.0 55 1.0
218 1.0 12 1.9
192 1.0 38 1.2
227 1.0 3 1.2
199 1.0 31 1.8
227 1.0 3 1.2
205 1.0 25 1.7
210 1.0 20 1.5
CIb
-- 1.24.6
-- 0.72.0
-- 0.71.6
-- 0.75.3
-- 0.71.9
-- 0.25.9
-- 1.03.3
-- 0.26.0
-- 0.93.4
-- 0.73.2
GARABRANT/BERRYMAN 00059
AEP Vol. 13, No. 7 August 2003: 485494
Adegoke et al. 491
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
TABLE 5. Odds ratios of leukemia by lifetime job occupational exposure by self-report combined with job-exposure matrix assessment among all participants
Self-report/JEM
Controls (n 502)
All leukemias (n 486)
Cases ORa
CIb
ALL (n 81)
Cases ORa
CIb
AML (n 236)
Cases ORa
CIb
CML (n 79)
Cases ORa
CIb
Benzene Neverc Everd
Other organic solvents Neverc Everd
EMF Neverc Everd
Pesticides Neverc Everd
457 45
410 92
355 147
474 28
420 65
392 93
356 129
460 25
aAll OR (odds ratio) adjusted for age, sex, and income. b95% confidence interval. cNo on both the self-report and the JEM. dYes on either self-report or JEM or both.
1.0 1.6
1.0 1.1
1.0 0.9
1.0 0.9
-- 1.12.4
-- 0.81.5
-- 0.71.2
-- 0.51.6
72 9
69 12
55 26
79 2
1.0 -- 210 1.3 0.62.9 26
1.0 -- 190 0.7 0.41.4 46
1.0 -- 179 1.1 0.61.8 57
1.0 -- 223 0.3 0.11.5 13
1.0 -- 1.3 0.82.2
1.0 -- 1.1 0.81.7
1.0 -- 0.8 0.61.2
1.0 -- 0.9 0.51.9
64 15
57 22
48 31
73 6
1.0 -- 2.4 1.34.7
1.0 -- 1.8 1.03.1
1.0 -- 1.6 1.02.7
1.0 -- 1.4 0.53.5
DISCUSSION
This population-based case-control study assessed the possible association of the risk of adult leukemia with occupational history and several occupational exposures using self-reported and JEM assessments among a population in Shanghai. It has a relatively large number of cases thus providing the opportunity to examine the relationship between occupational exposures to leukemia by its subtypes. Although the literature suggests that smoking may be a risk factor for leukemia (14), cigarette smoking was not associated with leukemia risk in this population and was not adjusted for in our data analyses.
The risk for leukemia was increased significantly among those who worked in the chemical manufacturing industry (OR = 3.1, 95% CI 1.09.8), consistent with findings from some (1921) but not all previous studies (22, 23). We noted borderline significant increases for risk of leukemia among textile workers and officials, leaders and managers. This is similar to the study by Costantini et al. (2) who reported a doubling in relative risk for leukemia among managers. Miligi et al. (24) had previously noted an increase in risk for leukemia among textile workers, as well as among housewives, teachers and hairdressers. We did not see any increase in risk among housewives and teachers. Several studies had earlier reported an increased risk of leukemia among nursing and healthcare workers (2527). We found an increased risk for leukemia among medical and public health workers, but it was not significant. The increase in risk for leukemia that we observed in primary metal workers, though non-significant, is consistent with what has been reported in an earlier study (27). The increase in risk for leukemia among construction workers that we observed has also been
previously reported (28). However, our own finding was not significant.
In analyses by specific substances, exposure to benzene, radioactive materials, synthetic fiber dust, and toluene all were associated with an elevated risk of leukemia. We also found that potential exposure to paints was associated with an increased risk for leukemias, but not significantly. Exposure to radioactive materials was more strongly associated with ALL, AML, and CML than exposure to benzene. Synthetic fiber dust was associated with more than two-fold risks for ALL, AML, and CML. A review of the literature revealed that very little has been reported about synthetic fiber dust as an independent risk factor for leukemia. In general, longer durations of exposure to benzene, radioactive materials, and toluene resulted in larger relative risks to all leukemias, ALL, AML, and CML than shorter durations. Wang et al. (7) had previously reported an increased risk to leukemia associated with increased duration of exposure to radioactive materials among Chinese workers. The risk of all leukemias combined following occupational exposure to toluene, and of CML following exposure to EMF, paints, and toluene also increased with increasing years of exposure. Seventy-eight percent of the subjects who reported exposure to toluene and 52% of the subjects that reported occupational exposure to paints also reported exposure to benzene. Industrial and environmental benzene is present as a contaminant or constituent of many industrial solvents (29, 30). Therefore, the risk related to occupational exposure to toluene and to paints may actually be due to the co-existence of benzene.
The combination of self-reports of exposure with assessments from the JEM, slightly increased the number of
GARABRANT/BERRYMAN 00060
492 Adegoke et al.
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
AEP Vol. 13, No. 7 August 2003: 485494
cases exposed to the various substances and yielded a statistically significant elevated risk for leukemia for benzene exposure. This combined exposure assessment also revealed significantly elevated risks for CML for benzene, other organic solvents, and EMF. These results were consistent with those from the self-report analysis. It has generally been accepted that benzene exposure in human populations is causally related to the development of ANLL (or AML) (31, 32). We observed a significant association between benzene exposure and CML. We noted a 60% excess risk for CML among subjects ever exposed to EMF by combined self-reported and JEM assessment, but results in the literature are mixed. A recent mortality study in the UK of electricity generation and transmission workers did not show an association between occupational exposure to EMF and the risk of leukemia (33), but a recent study by Minder and Pfluger (34) did.
We saw no clear gender differences for the risk for all leukemias combined for several occupational exposures, but some variation existed for others. The association between leukemia and pesticide exposure showed a deficit among men and an excess among women. On the other hand, the significant excess risk for leukemias following exposure to radioactive materials was only among men, but few women reported exposure to radioactive materials. The reason for this is unclear and it could be a chance finding, or genderrelated differences in exposure pattern (e.g., frequency, intensity, duration and mode of exposure) and host factors that affect chemical absorption and metabolism.
This study has several strengths. It is relatively large and allowed analyses by histologic type. It was conducted in the People's Republic of China, which has several advantages, including fewer jobs held per subject, and fewer exposures per subject (35). By holding fewer jobs, Chinese workers generally have fewer types of occupational exposures than workers in western industrialized countries; thus the potential for confounding by other hazardous substances is reduced. Studies in China also present an unusual opportunity to evaluate cancer risks from occupational exposures among women because they are typically employed in production jobs with exposures similar to those experienced by men. Evaluation of occupational risks among women can, thus, be done more effectively than in western countries, which is something desperately needed (36).
Since this study was an interview-based case-control study, there are several possible limitations. Recall bias is a concern because cases may have had a tendency to recall more factors that had been discussed with them in the past during their clinical management process than controls. The possible over-report of exposure or more accurate recalls by cases will bias the ORs away from the null. Nondifferential exposure misclassification is also a potential problem because of the high proportion of interviews obtained
from the next-of-kin of cases (48%). This would tend to bias the ORs towards the null. It is difficult to validate selfreported data on long-term exposure. However, we found in our study that results were similar between directly interviewed subjects and surrogate interview subjects and that self-reported assessment of occupational exposure appeared to be in a good agreement with JEM assessment in this population. For example, the ORs for all leukemia for "ever" exposed to benzene increased from 1.7 among all subjects to 2.3 among directly interviewed subjects only, both statistically significant, and for ALL increased from 1.5 among all subjects to 2.0 among directly interviewed subjects only, both non-significant. In addition, although the literature is not entirely consistent, some methodological studies have shown that self-reported occupational exposure could be useful in epidemiological studies especially when objective information on occupational exposure is not available (3742).
In addition, although the JEM used is specifically developed for occupations commonly held by the Chinese in Shanghai, it may not reflect the true occupational exposures. As with most observational epidemiologic studies of rare diseases, such as leukemia, the exposure assessment in this study was based on self-reported data. We recently compared self-reported data with those derived from a job-exposure matrix in this study population and found they are highly correlated. Agreement between self-reported exposures and JEM assessment was good (kappa coefficients []:0.480.84) (43). With JEM as the "gold standard", the sensitivities for the self-reported exposures ranged from 0.75 to 0.98. However, that for pesticide exposure was 0.44 in subjects 51 years of age. The self-reported exposures specificities ranged from 0.87 to 0.99 (43). We were unable to compare the JEM with actual environmental or biological measurements, as we did not have this information. It should be noted that despite the large size of this study, some of the results by subtype of leukemia were based on small numbers that may render the association estimates imprecise. Chance findings are sometimes unavoidable in epidemiologic studies when multiple exposures are evaluated, however, most of the associations reported in this manuscript are biologically plausible and consistent with results from in vitro, animal experiments, and limited epidemiologic studies conducted in other populations. Furthermore, the internal consistency of our findings, including the dose response relationships, also argues against the possibility of chance findings for most of the associations we have reported.
In summary, we found an association between risk of adult leukemia and working in the chemical industry. Analyses of specific exposures revealed increased risks for all leukemias from occupational exposure to benzene, synthetic fiber dust, radioactive materials, and toluene. Radioactive
GARABRANT/BERRYMAN 00061
AEP Vol. 13, No. 7 August 2003: 485494
Adegoke et al. 493
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
materials and synthetic fiber dust were identified as possible risk factors for ALL, AML, and CML specifically. Likewise, benzene, other organic solvents, EMF, and toluene were identified as possible risk factors for CML. The risk of all leukemias combined increased with duration of exposure to benzene, radioactive materials, and toluene, the risk of ALL rose with duration of exposure to radioactive materials, AML with duration of exposure to benzene, and CML with duration of exposure to benzene, EMF, radioactive materials, paints, and toluene. This implies that continuous long-term exposure rather than short-term exposure to these occupational factors increases the risk of workers developing leukemia.
This work was supported by a research grant from the Young Scientist Foundation, Chinese Ministry of Public Health. Dr. Adegoke is supported in part by a grant (1U54-CA9140801) from the National Cancer Institute and National Institutes of Health.
REFERENCES
1. Zeeb H, Blettner M. Adult leukaemia: what is the role of currently known risk factors? Radiat Environ Biophys. 1998;36:217228.
2. Costantini AS, Miligi L, Kriebel D, Ramazzotti V, Rodella S, Scarpi E et al. A multi-center case-control study in Italy on hematolymphopoietic neoplasms and occupation. Epidemiology. 2001;12:7887.
3. Sathiakumar N, Delzell E, Cole P, Brill I, Frisch J, Spivey G. A casecontrol study of leukemia among petroleum workers. J Occup Environ Med. 1995;37:12691277.
4. Schnatter AR, Armstrong TW, Thompson LS, Nicolich MJ, Katz AM, Huebner WW, et al. The relationship between low-level benzene exposure and leukemia in Canadian petroleum distribution workers. Environ Health Perspect. 1996;104(Suppl 6):13751379.
5. Nilsson RI, Nordlinder R, Horte LG, Jarvholm B. Leukaemia, lymphoma, and multiple myeloma in seamen on tankers. Occup Environ Med. 1998;55:517521.
6. Lazarov D, Waldron HA, Pejin D. Acute myeloid leukaemia and exposure to organic solvents--a case-control study. Eur J Epidemiol. 2000;16:295301.
7. Wang J, Boice JD Jr, Inskip PD, Li B, Zhang J, Fraumeni JF Jr. Leukemias among medical diagnostic X-ray workers in China. Proc Chin Acad Med Sci Peking Union Med Coll. 1990;5:194199.
8. Hayes RB, Yin SN, Dosemeci M, Li GL, Wacholder S, Travis LB et al. Benzene and the dose-related incidence of hematologic neoplasms in China. Chinese Academy of Preventive MedicineNational Cancer Institute Benzene Study Group. J Natl Cancer Inst. 1997;89:10651071.
9. Wong O, Raabe GK. Cell-type-specific leukemia analyses in a combined cohort of more than 208,000 petroleum workers in the United States and the UK, 19371989. Regul Toxicol Pharmacol. 1995;21:307321.
10. Parkin DM, Whelan SL, Ferlay J, Raymond L, Young J, eds. Cancer Incidence in Five Continents. v. 7. IARC Scientific Publication. Lyon: IARC; 1997:143.
11. Ferlay J, Parkin DM, Pisani P. GLOBOCAN 1: Cancer Incidence and Mortality Worldwide [CD-ROM]. Lyon: IARC Press; 1998.
12. Zheng W, Linet MS, Shu XO, Pan RP, Gao YT, Fraumeni JF Jr. Prior medical conditions and the risk of adult leukemia in Shanghai, People's Republic of China. Cancer Causes Control. 1993;4:361368.
13. Groves FD, Linet MS, Devesa SS. Epidemiology of human leukemia. Curr Opin Hematol. 1994;1:321326.
14. Linet MS, Cartwright RA. The Leukemias. In: Schottenfeld D, Fraumeni JF Jr, eds. Cancer Epidemiology and Prevention. New York: Oxford University Press; 1996;841892.
15. Yuasa H, Hamajima N, Ueda R, Ohno R, Asou N, Utsunomiya A, et al. Case-control study of leukemia and diagnostic radiation exposure. Int J Hematol. 1997;65:251261.
16. Gerin M, Siemiatycki J, Desy M, Krewski D. Associations between several sites of cancer and occupational exposure to benzene, toluene, xylene, and styrene: results of a case-control study in Montreal. Am J Ind Med. 1998;34:144156.
17. Levin LI, Zheng W, Blot WJ, Gao YT, Fraumeni JF Jr. Occupation and lung cancer in Shanghai: a case-control study. Br J Ind Med. 1988;45:450458.
18. Breslow NE, Day NE. Statistical Methods in Cancer Research. v. 1. The analysis of case-control studies. IARC Scientific Publication. Lyon: IARC; 1980; 32.
19. Leet T, Acquavella J, Lynch C, Anne M, Weiss NS, Vaughan T, et al. Cancer incidence among alachlor manufacturing workers. Am J Ind Med. 1996;30:300306.
20. Weimer K, Talbott EO, Day RD. Further examination of employment in the chemical industry: follow-up case-control study of hematopoietic and lymphoid neoplasms using a next consecutive death control group. Am J Ind Med. 1998;33:9798 [Letter].
21. Massoudi BL, Talbott EO, Day RD, Swerdlow SH, Marsh GM, Kuller LH. A case-control study of hematopoietic and lymphoid neoplasms: the role of work in the chemical industry. Am J Ind Med. 1997;31:2127.
22. Burns CJ, McLaury RL, Jammer BL, Cartmill JB. Mortality experience for Louisiana chemical workers. The Journal of the Louisiana State Medical Society. 1999;151:4246.
23. Tsai SP, Gilstrap EL, Colangelo TA, Menard AK, Ross CE. A mortality study of oil refinery and petrochemical employees. J Occup Environ Med. 1997;39:448454.
24. Miligi L, Costantini AS, Crosignani P, Fontana A, Masala G, Nanni O, et al. Occupational, environmental, and life-style factors associated with the risk of hematolymphopoietic malignancies in women. Am J Ind Med. 1999;36:6069.
25. Peipins LA, Burnett C, Alterman T, Lalich N. Mortality patterns among female nurses: a 27-state study, 19841990. Am J Public Health. 1997; 87:15391543.
26. Petralia SA, Dosemeci M, Adams EE, Zahm SH. Cancer mortality among women employed in healthcare occupations in 24 US states, 19841993. Am J Ind Med. 1999;36:159165.
27. Blair A, Zheng T, Linos A, Stewart PA, Zhang YW, Cantor KP. Occupation and leukemia: a population-based case-control study in Iowa and Minnesota. Am J Ind Med. 2001;40:314.
28. Robinson CF, Petersen M, Palu S. Mortality patterns among electrical workers employed in the US construction industry, 19821987. Am J Ind Med. 1999;36:630637.
29. Linet MS, Yin SN, Travis LB, Li CY, Zhang ZN, Li DG, et al. Clinical features of hematopoietic malignancies and related disorders among benzene-exposed workers in China. Environ Health Perspect. 1996;104(Suppl 6):13531364.
30. Wester RC, Maibach HI. Benzene percutaneous absorption: dermal exposure relative to other benzene sources. Int J Occup Environ Health. 2000;6:122126.
31. Savitz DA, Andrews KW. Review of epidemiologic evidence on benzene and lymphatic and hematopoietic cancers. Am J Ind Med. 1997;31:287295.
32. Hayes RB, Yin S, Rothman N, Dosemeci M, Li G, Travis LT, et al. Benzene and lymphohematopoietic malignancies in China. J Toxicol Env Heal A. 2000;61:419432.
33. Harrington JM, Nichols L, Sorahan T, van Tongeren M. Leukaemia mortality in relation to magnetic field exposure: findings from a study of UK electricity generation and transmission workers, 197397. Occup Environ Med. 2001;58:307314.
GARABRANT/BERRYMAN 00062
494 Adegoke et al.
OCCUPATIONAL EXPOSURE AND LEUKEMIA IN SHANGHAI
AEP Vol. 13, No. 7 August 2003: 485494
34. Minder CE, Pfluger DH. Leukemia, brain tumors, and exposure to extremely low frequency electromagnetic fields in Swiss railway employees. Am J Epidemiol. 2001;153:825835.
35. Dosemeci M, Cocco P, Gomez M, Stewart PA, Heineman EF. Effects of three features of a job-exposure matrix on risk estimates. Epidemiology. 1994;5:124127.
36. Blair A, Zahm SH, Silverman DT. Occupational cancer among women: research status and methodologic considerations. Am J Ind Med. 1999; 36:617.
37. Van der Gulden JWJ, Jansen IW, Verbeek ALM, Kolk JJ. Repeatability of self-reported data on occupational exposure to particular compounds. Int J Epidemiol. 1993;22:284287.
38. Holmes E, Garshick E. The reproducibility of the self-report of occupational exposure to asbestos and dust. J Occup Med. 1991;33:134138.
39. Engel LS, Seixas NS, Keifer MC, Longstreth WT Jr, Checkoway H. Validity study of self-reported pesticide exposure among orchardists. J Expo Anal Environ Epidemiol. 2001;11:359368.
40. Cordeiro R. Validity of occupational histories from proxy respondents. Rev Saude Publica. 2000;34:522528 [Abstract].
41. Fritschi L, Siemiatycki J, Richardson L. Self-assessed versus expert-assessed occupational exposures. Am J Epidemiol. 1996;144:521527.
42. Arbuckle TE, Burnett R, Cole D, Teschke K, Dosemeci M, Bancej C, et al. Predictors of herbicide exposure in farm applicators. Int Arch Occup Environ Health. 2002;75:406414.
43. Adegoke OJ, Blair A, Shu XO, Sanderson M, Addy CL, Dosemeci M, et al. Agreement of job-exposure matrix (JEM) assessed exposure and selfreported exposure among adult leukemia patients and controls in Shanghai. American Journal of Industrial Medicine. In press.
GARABRANT/BERRYMAN 00063