Document qaexpzMJbL668R9XL80qrOGQE

~-~_... .o Ketrospective Mortality and Medical Surveillance Studies of Workers Benzene Areas of Refineries Shan P. Tsai , Ph. joC. P. Wen , M. ; Nancy S. Weiss , M. and Roy L. Gibson , M. ; Otto Wong, Sc. ; William A. McClellan , M. The mortality and health experiences of refinery work- ers employed in benzene processes or operations are de- scribed. A retrospective cohort mortality study of benzene workers employed from 7952 to 7978 revealed no excess in overall general mortality or in cancer mortality compared either with the experience of the U. S. general population or with that of an internal control group. Ascertainment of vital status was accomplished for 99% of the cohort. Recent industrial hygiene data that included 394 personal sam- ples indicated that 84% of all benzene exposures were less tl-. part per million (ppm), with a mrdian exposure of flpm for the refinery workers, and 0. 53 ppm for those in the benzene-related units. Among these workers, no deaths from leukemia were observed. A medical surveillance program for benzene workers is also described, with special emphasis on the effectiveness of laboratory screening. E valuation of data for a 27 year period showed no signifi- cant changes in the blood indices of the workers as a group. The limited value of establishing screening guidelines with- out the support of epidemiological studies is discussed. The potential effects on workers of chronic benzene exposure is a subject of increasing attention. l,2 Animal and human studies have focused on potential insults by benzene to the hematopoietic system. Numerous case reports and several epidemiological studies have attempted to establish the leukemogenicity of benzene. B The results , however have not been consistent, and data on human health effects at the exposure level of current operations of petrochemical industries are limited. Due to the ubiquitous nature of benzene in many industries , the control of its hazards will continue to be a complex issue. This study examines the mortality and health experiences of refinery workers employed in benzene processes or operations. Of all the employees at the refinery, the work- ')m the Gulf Oil Corporation (Ms. Weiss and Drs. Tsai , Wen ~ Ian , and Gibson); and Environmental Health Associates Be, ~"Iey, Calif. (Dr. Wong). Address correspondence to Gulf Oil Corporation , P. O. Box 2100 , Houston , TX 77252 (Dr. Wen). exceeding seven hours; partial samples are those with samp- exposure and would be the ones most Ii kely to demonstrate adverse mortality and health patterns. In addition to exposures to benzene , these workers have also had potential exposures to other chemicals found in a typical refinery setti ng. This report is divided into two parts. The first part describes a retrospective mortality study of benzene workers employed from 1952 to 1978 at a Texas refinery. The second part is an evaluation of a 21-year medical surveillance system for benzene workers in place since 1959 at the same refi nery. I. RETROSPECTIVE MORTALITY STUDY OF BENZENE WORKERS Materials and Methods Definition of the Benzene Cohort - The study includes all male workers ever employed at a large Texas refinery , 1978from Sept. 15 , 1952 , to Jan. 1 (regardless of the length of their employment) who worked directly on the benzene , ethylene , aromatic distillate hydrogenation (ADH) or cumene units which are the principal petrochemical units in the refinery. Maintenance workers assigned on a regular basis to the benzene-related units , utility men , and laborers assigned to these units were included in the cohort. However, maintenance workers who were assigned to benzene units on an irregular basis were not included. The ethylene unit began operation of Sept. 15 , 1952; benzene on Sept. 14 , 1959; ADH on May 30 , 1962; and cumene on Jan. 5 , 1970. Exposure Data - Data of exposure to benzene for the years 1973 through 1982 were examined. Data for the years prior to 1973 were inadequate for assessing the work- ers ' exposure. The personal sampling exposure data for the 10-year period have been compiled for employees from the entire refinery and are shown in Table 1. Benzene was measured by the silica gel or midget im- pinger method at an earlier time period and by charcoal tubes later. Data were divided into two categories according to the sampling time. Full samples are those with sampling time greater than or equal to four hours, with the majority Journal of Occupational Medicine/Vol. 25 , No. 9/September 1983 685 GARABRANT/BERRYMAN 00354 ..tiii Table 1 - Frequency Distribution of Benzene Exposure for Refinery Workers, 1973.1982 Benzene Concentration ppm Full No. of Samples (%) Partial No. of Samples (%) Combined No. of Samples (%) -::::0. 1.0- 10. 24. ;;.25. 378 (52) 270 (37) 66 (9) 10 (1) 1 (0) 2 (0) 222 (33) 307 (46) 97 (14) 19 (3) 12 (2) 13 (2) 600 (43) 574 (41) 163 (12) 29 (2) 13 (1) 15 (1) Total 727 (99) 667 (100) 394 (100) ers in this operation have presumably received the greatest ling time less than four hours or those recorded as partial samples whose data on sampling time were missing. No attempt has been made to adjust the exposure numbers to eight- hour time-weighted averages (TWAs). The sampling strategy used by industrial hygienists for determining personal exposure has usually been based on representative sampling of various job assignments or worst case sampling on spills or turnaround. Job classifica- tions selected for sampling are usually those in which the highest potential for exposure exists during routine opera tions. Replications , of these job classifications are then sampled to characterize the exposures. Data Collection - Copies of all personnel records since the refinery operation was started in 1901 were made , and pertinent demographic and work history information were screened according to the cohort definition previously described. Of the approximately 20 000 employees whose records were collected and screened , 454 have been identified as benzene workers. Data collected on each cohort member included name , date of birth , race , sex , social security number , date of hire , employment status , date of last separation , if any; vital status; date of death for known decedents and cause of death; and work history. Each work history consisted of a chronological listing of all jobs ever held by the individual and included the date of job assignment department, unit and job title. Standard epidemiologic procedures for ascertaining the vital status of the cohort were followed. They included matching the cohort employees with rosters of current employees and annuitants , matching with death certificate files of company records , obtaining information from the Social Security Administration , and the Texas and Louisiana Departments of Motor Vehicles and a plant follow-up. Death certificates were collected from the vital statistics departments of individual states for all known decedents. The underlying cause of death was coded by a trained nosologist according to the rules of the International Classification of Diseases (ICD) revision in effect at the. time of death and then recoded to the ICD adapted (ICDA) eighth revision. F or analytical purposes , numbers of person-years at risk were calculated. The denominator data were determined from Sept. 15 , 1952 , or from the date of first benzene exposure , whichever was later , to the end of the study period (J an. 1 , 1978) or to the date of death , which ever was earlier. Individuals with unknown vital status were assumed lost to follow-up on the date of termination or on the last date of contact , whichever was later. The number of years contributed by each worker was classified by age race , and calendar year. The U. S. national age-sex-race-yearcause-specific mortality rates for five-year time intervals during the study period were applied to these numbers of person-years to obtain the number of expected deaths from a specific cause. Standardized mortality ratios (SMRs) were computed and their significance testing is based on the assumption of a Poisson distribution for the observed numbers of deaths , utilizing a two-sided test of significance. Part of the analysis was performed using a computer pro- gram developed by Monson. An internal comparison group consisting of a 10% sample of the employees from the same plant who worked during the same periods in operations not related to benzene was selected. The selection guide was social security number and the analysis was based on relative risks deter- mined by using the Mantel- Haenszel procedure. Results Table 2 shows the characteristics of the benzene cohort. Of the total cohort of 454 employees , 369 (81 %) were white and contributed a combined total of 5 900 personyears. The average age of study subjects at entry into the cohort was approximately 34 years and the average length of follow-up 13 years. The average number of years of benzene-related work (exposure) was 8. 0 for the white employees and 4. 5 for the nonwhite employees. The cohort is relatively young with only 7. 5% (N=34) of the employees found to be deceased. All death certificates were identified and only three employees were lost to follow-up. Thus , the follow-up was successful for 99. 3% of the cohort. Distributions of the benzene cohort by year of fi rst exposure and duration of exposure are shown in Tables 3 and 4 , respectively. One-half of the cohort began work in benzene areas before 1965 and almost 80% had exposures lasting one year or longer. Table 2 - Benzene Cohort Statistics Sept. 15, 1952 to Jan. 1 , 1978 Characteristics White Nonwhite No. studied Person-years of observation Average age at entry Average follow-up, yr Average exposure , yr No. alive No. dead Death certificate unavailable No. lost to follow-up 369 (81) 000. (85) 33. 13. 340 (0. (19) 894. (15) 37. 10. Total 454 900 33. 13. 417 (0. 686 Refinery Workers and Benzene/Tsai et al GARABRANT/BERRYMAN 00355 - .-- ----::...-- - - .-. ------ Table 3 - Distribution of Cohort by Year of First Exposure Year of First Exposure 1952- 1954 1955- 1959 1960- 1964 1965- 1969 1970- 1974 1975- 1977 Total No. (%) 61 03.4) 102122. 65 04. 7306. 104122. 49 (10. 454 000. (50. Table 4 - Distribution of Cohort by Duration of Exposure Length of Exposure, yr -::1 1-4 1015- ;.20 Total No. (%) 95 120. 145 (31.9) 77 07. 52 (11.4! 47 (10. 38 (8. 454 (100. The total number of samples of personal exposure to benzene obtained there were two or mwaosre1sa7m18p.lesAvferormagethsewsearme etaken when on the shown same day, resulting in 1 in Table '1. For 89% of 394 samples for the full samples individuals analysis and , as the partial samples exposures were less than 1 pa7r9t%peorf million (ppm) benzene. For 10% of the full 17% of the partial samples samples and 10 ppm. In addition exposures were between 1 and , for 4% of the partial samples expo- sures exceeded 10 ppm. Table 5 shows benzene exposures by selected units. As tehxopseectaesds,igthneedmtoedthiaencbuemnezneeneanexdpboesnuzreenwe aasnhdicgyhcelsothaemxaonneg units. This was followed by amine , waste water treatment dpeusmcepnhdoinugseo,rldaebrobraastoedryo, netfhuylllesnaem,pAleDs.HFuonr ipt,aratniadl dsaomckpsl,eisn mdthoeecnkots,r,daAenrdDwpHausmucpnumiht,oenulesae,b.obrSeanitnzocerenye,thaaenmdsitncueyd,cywlodahesestxiegannwewa,ateestrhliytmrleeitaneetd- to those working in the benzene and cyclohexane ADH , and ethylene units , these units were , grouped cumene in Table 6 as listed "inbeTnazbelnee5-rewlaetreedculansitssif"iedwhaisle" otthheer remaining units benzene areas. All of the remaining samples taken in the crelalassteifdieudnaitss"walelroeththeersh. i"ghTheest ewxpitohsuarmesedfiraonm refinery were otfhe0. benzene- followed by those from other benzene areas with 0. 53 ppm Exposures from all others were clearly the lowes2t 4wpitphma. median of 0. 07 ppm. Table 7 presents the'SMRs for selected causes of death for all males (white and nonwhite combined). The cohort as a whole showed a favorable overall mortality There were a total of 34 deaths pected. The expected numbers of dwietahth5s8w. 6e9reedxepaetrhisenecxe-. for white and nonwhite males independently basceadlcounlattehde S. white and nonwhite male mortality rates and then were added together. The SMR for all significantly decreased with arteriosclerotic heart disease causes (0. 58) is 01. The mortality for showed a similar significant deficit (SMR=0.46). The number of cancer deaths (10) was slightly smaller than expected (11.53; SMR=0. were no cases of leukemia or of cancer of the 8l7y)m.pThhateirec system. Excesses were seen for cancers of the stomach intestine and brain , each with two deaths. Cancer o,fltahrege buccal cavity and pharynx and cancer of the larynx were also increased , each with one death. The numbers were too small to permit any conclusion and chance variation could explain the excesses. However, this points out the need for continuation of the surveillance of benzene workers to verify the significance observations. , or the lack of significance , of these Table 8 shows SMRs for those who worked one year or more in the benzene areas. The total number of the cohort decreased from 454 to 359. The results are shown in Table 7. The total number of cansicmeirlar to those creased from 10 to four and the SMR for deaths de- from 0. 87 to 0.49. cancer changed Table 5 - Benzene Exposure Concentrations (ppm) by Selected Units* Full Unit Median Mean, SO GM 1 GSO Benzene and cyclohexane Cumene AOH Ethylene Pump house No. 11 Docks 'TIine ,boratory Wastewater treatment 1.341 1.39 801 1.18 10,10. 1.37 1 331 1.01 041 11, 071 551 1.12 12115. 10: 061 1.25: 1.24 641 92: 06114. 1.16: 371 . SO indicates standard deviation; GM , geometric mean; GSD , geometric standard deviation Median 1.25 1.60 Journal of Occupational MedicinelVol. 25 , No. 9/September 1983 Partial Mean 1 SO 561 961 14.42 741 27, 45123. 651 491 051 411 GM 1 GSO 1.441 1.851 08116. 11 1 12113. 551 401 491 28, 687 GARABRANT/BERRYMAN 00356 GARABRANT/BERRYMAN 00357 GARABRANT/BERRYMAN 00358 GARABRANT/BERRYMAN 00359 GARABRANT/BERRYMAN 00360 GARABRANT/BERRYMAN 00361