Document MJxaRy38awqnkkZa34LEa5vBj

Epidemiologic Survey June 1974 Vol. 16 No. 6 of Leukemia in Persons Potentially Exposed to Benzene John J. Thorp, M.D. A review of the literature on the epidemiology of leukemia i n persons exposed to benzene reveals a lack of broad based population studies. Browning' in 1965 reported a total of 61 cases of various types which had been reported between 1897 and the early 1960's. The majority of these were single cases, with seven being the largest number in any one series. The lack of data on the 1897 case makes the diagnosis doubtful and according to Dameshek and Gum* the first clearly established case of leukemia in a benzol worker was published in 1928 and the second in 1932. Vigliani and Saita3 in 1964 reported that they had seen 47 cases of benzene hemopathy between 1942 and 1963 at the Clinica -del Lavoro six of these had leukemia During the period 1960 to I963 the neighboring Institute of Occupational Health in Pavia observed five cases of leukemia. All the rest of Italy reported 13 cases during the period 1941 to 1963. The majority of these were workers in small plants or cottage industries where solvents or glues containing up to 60% benzene were used and the exposures were apparently high, although no exact measurements were made. The duration of exposure ranged trom 4 to 20 years. Vigliani also cites Cavignaux' who reported on 42 cases of benzene myelopathy occurring in France during the years 1960 to 1961, six of these were cases of leukemia. Aksoyj in 1972 reported the deaths trom leukemia of four Turkish shoemakers who worked for periods of 6 to 14 years i n unhygienic conditions where air concentrations between 150 and 210 ppm were measured. Although the fact of excessive benzene exposure appeared evident in all of these cases there was little, if any, indication of the total number of workers at risk In an earlier work by the same group of investigator+ however, they had reported a study which conforms more closely to a true epidemiological investigation. They compared the hematological findings in a control group of 1 0 0 males with those in 217 apparently healthy males employed in a number of poorly ventilated s h o e shops where benzene solvents were used. Benzene exposures ranged between 30 and 210 ppm for periods of three months to 17 years. Although 51 (23.5%) of the exposed workers showed hematological abnormalities, consisting of leucopenia, thrombocytopenia or pancytopenia, no cases of leukemia were observed. Mild to moderate hypochromic or normochromic anemia was present in the exposed group but was also noted in 21%of the controls, so the authors were hesitant about ascribing this abnormality solely to benzene. One other fairly large scale study in the field was reported in 1971 by lshimaru et a17 as part of the follow-up work done in lapan by the Atomic Bomb Casualty Commission. The major work oi this group has been related to the correlation of the radiation dose received and the development of pathological changes i n various organ systems. In particular their work has shown that an Journal of Occupational YedicineNol. 16, No. 6/June 1974 increase in leukemia incidence can be related to a single dose of 1 0 0 RADS or more, that the modal incubation time is about five years and that approximately 90% of cases will occur within ten years of the exposure. In the work referred to here however, lshimaru and his co-workers attempted to evaluate occupational and other environmental factors in relation to the incidence of leukemia in persons exposed itto the radiation from single atomic bomb blasts in Hiroshima and Nagasaki in 71945. In 303 pairs of a ult leukemia cases and controls, the risk was ap- proximately 2.5 times higher among those with a history of probable OCcupational exposure to benzene or its derivatives and medical x-ray. Unfortunately, the retrospective nature of the study, the fact that the(environmenta1 exposure data in many cases had to be obtained from relatives or close friends of the leukemia victims and a lumping together of occupations where solvents probably containing bendene were used with those where medicalx-ray exposure existed, make it impossible to really isolate the benzene effedt. This somewhat confusing situation i s not clarified by the fact tqat experimental attempts to induce leukymia in aliimals have not been uniforhly successful.8 Thus, while there are a qumber of cases where exposure to high, levels of bmzene i s followed by marked hematopoetic changes, including acute leukemia, there i s less than solid scientific evidence for attributing leukemia to exposure to low levels of benzene.8 The comments by an official of a European government agency that data 375 .- from medical records of a company such I - as Exxon might influence their action in recommending a ceiling limit on ben- zene provided the incentive for un- dertaking this study. Eight European affiliates were asked to participate and all responded generously and en- hliu WUdiOn Phat ornpliw plrnt r m p l i q Tank dipping Winhr Summer Summer thusiastically to our request for assistance. Methodology The eight affiliates chosen were select- Buum wad Plant rmptim Tank dipping Dcanmng vmtm tun hnk Sum1 Sumr Summer (0.0s to 0.5 3A -N c'4 5* ed because past experience had in- reasons it was decided to confine the filiate for the 10-year period 1962 to 1971 dicated they had a fairly broad based ex- study to leukemia. chosen for study. This gave us a perience with the manufacture, trans- Certain methodological problems reasonably large denominator for portation and marketing of petroleum were recognized from the onset of calculating incidence rates. We were or chemical products containing ben- the study. They were: (1) Low incidence able to obtain the total employe zene for at least ten years. They also had of leukemia in the general population; population figure from all of the affiliates well organized periodic examination and (2) The validity of the diagnosis of the for the years in question and the an- absentee data collection for their entire cases of leukemia reported; (3) The nuitant totals from six of the eight. The population for a similar period. quantitative definition of the extent of age breakdown for the entire population The participating affiliates were: (1) exposure to benzene of the population for each year of the IO-year period was AIS Norske Esso; (2) Esso A.C.; (3) Esso under study; (4) The inadequacy of unobtainable. We were, however, able Belgium S.A.; (4) Esso Italiana; (5) Esso follow-up on annuitants; and ( 5 ) In- to obtain this data on employees for Nederland N.V.; ( 6 ) Esso Petroleum Co. complete occupational histories on in- 1971 from Esso Europe. This information Ltd.; (7) Esso Standard S.A.F.; and ( 8 ) dividuals with the diagnosis of leukemia. plus the use of the percentage age Esso (Switzerland). The problems listed above may be distribution from general census data for Our original proposal was purposely amplified as follows: 1971 applied to the annuitant person rather general ir! natwe, particularly with 1. The incidence of leukemia in the years yielded an age distribution respect to the diagnostic categories in- general population w i t h an age population which we believe to be cluded. Figure 1 shows the form sent to distribution similar to that under study is reasonably accurate. the cooperating affiliates: very low - about three to eight cases 2. The validity of the diagnosis of As the completed forms were returned per 100,OOO population. This makes leukemia. The majority of the cases of it became obvious that entities such as aplastic anemia were not frequent enough to have any statistical validity or the diagnostic terminology did not permit comparison with statistics reported for the general population. For these meaningful evaluation of the occurrence of an odd case or two in a small company or unit difficult. To overcome this problem we have used the "personyears" concept. This involved cumulating the population of each af- leukemia were identified by a review of the annual absence data (which also includes deaths). Many of these were under the care of their own physicians and we had no means of verifying the laboratory or necropsy data which form- -Fir 1. Leukemia cpidemidgial study. ed the basis for the diagnosis. 3. The quantitative definition of the ex- tent of exposure to benzene of the 1RANGE OF X EINZENE TOTAL CO POP TOTAL NO OF CASIS'OF NO Of EYPL L ANNUITANTS 3 OF CASES of EL001 I~BIML)IWPRODUCTSI~KE~Yl~REI(EYPL LAWWUITANTS) WITHEXPOSURE TOMORE rSWASlASi lEYPL HANDLED IN- THAW I%OF BEWZEWE FOR ANMUITANTSWITH SYEARSIN. (POSURE TO MORE HAW I% OF BENE- EWE FORSYRS population under study is admittedly inexact. While data exist regarding certain jobs primarily related to the handling of benzene containing products (operators in benzene extraction plants, = barge operators, tank truck operators, 0 cc etc.) these are necessarily "spot" o b az servations. Shewood9 in 1971 sum- -vaacE) c marized the results of personal air sampling from routine operations in an aromatics plant (Table 1). In 1972 the same author10 reported ob- 1964 servations on two loaders and one weigher at a loading rack where petrol containing 2.7% and 4.5% benzene by weight was handled (Table 2 ) . 1970 1911 PI& M6 II An Institute of Petroleum study group headed by C. S. Parkinson reported results of a survey of retail filling stations 316 Epidemiological Survey of Leukemia and Potential Benzene ExposureRhorpe I1 -Table 3. Leukemia Epidcmiolqical Study Distribution of Cases ~ I -1 of Leukemia by Affiliate Essa AG. Esro Petroleum Co Ltd. Esso Standard S.AF Essa (Switzerland) I 3 a 6 1 I and normal gasoline transport operations in 1971.11 The survey which was carried out in August and September 1969 included nine filling stations, one road car loading station studied for two days, two rail car loading stations studied for six days and the loading of benzene enriched gasolines for three days at each of three locations. The following con- people with a potential exposure to 1% or more benzene over a 5-year period. Another way in which we approached the problem was to solicit available information on the percentage of benzene in products handled in the major func- tional units of each affiliate. From this we derived a "benzene exposure index" which gives a very crude exposure com- leukemia. Although the last job category of most of the cases of leukemia reported was available, a number had inadequate histories of what jobs they had held in the past or if they had any benzene exposure during a significant part of their working career. ~~~~l~~ I ! I' I 1 ! !i loading installations did not indicate any significant exposure to benzene." Until the 1950s employees tended to remain in specific job assignments in given locations in a plant for many years. With the introduction of manpower reduction and efficiency improvement programs in many company locations the concept of mobility or flexibility of job assignment developed. In some instances process operators did unit mechanical work and specific craft assignments gave way to the refinery mechanic concept, so that employees might work near a pump or line containing benzene one or two days and then might be in a benzene f r e e area for days or weeks. Only in locations where highly concentrated benzene was manufactured or transported are assignments reasonably constant and medical controls in effect. Our approach to this question was to - b wTrLk 4. Lrhemmir Epidrairkdal Study- knuuv d 18 eP.,d leukwh he at Death (or lnl),last bb Assignmenl and Trpr d Lukda. k ApatDrth ladlab- TjpadLwbk F 19 Ckk. He8d me L a b i a . Type? I 26 Engilwa, Rdinery L a k m i r Type? Y t 33' P a r c o w Car O r k (Tank TmcC. Driw 196066) C h i c Myeloid L a k r n i i Y: 36 k h m i c in R&uy L a k m i r Type? Y t 40 plwa Opuclta.Rdlm~ Y t 41 Fm. R&eq Myeloid Leuknia h t e Ieukunir Y 42 K u m e Truck Driver. Airport Llukmia. Typ? Y 46 Lthmticiul Head ok Leukemic. Type? Y t 60 Marketing SuprinimkRt Lakmia. Type? M 61 Industri~I Assistant k k L r t m i a , Type? Y 62' CapmW. R6w Lukmia. Type? I t 66 Truck O r i i . MarketiLctin( Lahemia, Type? Y sa C a m r a d Ckh Hpd Oftie 1u*mir b e ? M 68' Clerk Chwfirur. bhuy Lrkmic. Type? Y 69 n a . H a d M a C k m M~ ydotd L r * n i 8 Y t 70 Pmsrs Opasta. ReF~nuy Lahl?li8, Type? Y 71 Lnyr. Had rn Aolh Yplablrrtic Lmlrcnir It 73 * Mechanic. Mar*tin( Repaif Shs, C h i c L~npbtiCLuMr c ! jI 115-19 120-24 EXPOSED I5-29130-34135-39?0-44/45-49(50-54 5.59 0-64 55-89 60 0-j6 I vincing The number of persons 0 ~ casionally exposed 10 benzene or exposed to verv low benzene concentrations was s~ high that some cases of chronic leukemia could have cxcurred among them as among any other working population If the data were available and Vigliani - NOT EXPOSED 8 6 -STIFL A 1 'E IN - 0 68 0 69 is correct we might have had another parameter available by which to evaluate the relationship of the disease to possible exposure. Turning to the comparison of the experience of the eight affiliates in the study with that of the general population in the same geographical area, Table 7 shows the cumulative employee plus annuitant man-years and the number of cases of leukemia reported. -Fig 2. Age distribution of 18 azcs of leukemia (age at death or Der 1971). To make a valid comparison of the leukemia mortality experience between type of leukemia is given in Table 4. Figure 2 shows the cases distributed by 5-year age groups in the two maior subdivisions "Exposed" and "Not Exposed". As previously noted under Methodology we defined the "exposed" group as those individuals who worked for five or m.ore years in jobs where there was a potential exposure to 1%or and the kerosene truck driver as exposed. It so happened that the physician reporting these cases could categorize them as not having had any benzol exposure. An attempt to analyze the reported cases of leukemia b y various pathological types was unsuccessful as shown in Table 6. a population of 383,276 and that of the eight European countries involved with a total population many times greater, we chose to compare the "observed" with the "expected" number of deaths. This required apportioning the man-years experience into the same 5-year age classes used to report WHO mortality data. This could be done for the age groups 1 5 more benzene. While w e received numbers for this category from all affiliates, the physicians who reported them were well aware that this data was not as exact as it might appear. Some indicated that they were reasonably confident that the jobs included in this category did have an associated benzene exposure, usually at a low level, but the fact that one employee stayed in a particular job five or more years was more difficult to establish. Other medical directors indicated that the number reported in the "exposed" category was more indirectly derived: The number of people in administrative!clericaI lobs in headquarters or in the plants was subtracted from the total population and the difference was considered to represent the "exposed" group, What we have in Figure 2 then is, in effect, probably a spectrum extending from an occasional benzene exposure of unknown degree and duration down to absolutely no exposure. Table 5 recapitulates the jobs considered to be in the "Exposed" and "Not Exposed" categories and reinforces what has just been said about the difficulty in making a sharp distinction between the two. Under the indirect method of calculating exposure one w o u l d probably consider the refinery carpenter Iw M fQad M r* I ~~~ Tank IN&Driver Retimy W a n e Retwi Operata YrMing R q m Mechanic Markting Supsrintcndenl lkd toa 2 2 2 1 1 An Jab liik 33. 66 Clerk. Head OCa 36. 41 Manager, Head O f f i a 40. 70 Refinery Enoneer 73 Hathemtiaan. Head O f f i a 70 Rrfnery Carpenter Rcfnery Clerk Indl Assistant Dala Kerosene Truck Driver, Airowl lkd Clvr - AD 19. 68. 69 71 26 46 62 68 61 b2 While understandable in a retrospective study reaching back more than ten years, it i s unfortunate that w e can only be specific about 3 of the 18 cases Vigliani) has said We have never seen a case or chronic myeloid or lymphatic leukemia in workers poisoned with benzene all cases seen by us and at the Institute of Pavia in which harmful and prolonged exposure to benzene vapors was ascertained, were hemocytoblastic in type Whether chronic types of leukemia can be induced by benzene is still an open question, but w e must emphasize our view that in some of these cases reported in the literature the occupational history was not con- through 69 by multiplying the total employee man-vears (321,840) by the percentage distribution in each 5-year class reported to u s by Esso Europe tor 1971 -Table 6. Types 01 Leukemia Reported T r p I(. Chronic Lymphatic Leukemia Chronic Myelold Leukemia Myelold Leukemia Acute Myeloblastic Leukemia Acute Leukemia Leukemia, Type Unspecified lU 2' IU I In 12 In ''Expored'' Group Epidemiological Survey of Leukemia and Potential Benzene Exposure/Thorp When we sought a similar allocation of the 58,436 annuitant man-years we did not have an exact percentage distribution by age groups available within the region so we had to resort to an indirect approach. Utilizing the population of the U.K. (where 37% of our employee population is located) as recorded in the U.N. Demographic Year Book of 1971 for the 5-year age classes 55 through 85 and over, we applied the proportion of the general population in those age groups to our annuitant total. Table 8 shows the estimated age adjusted population arrived at after the above calculations. The "expected" number of deaths was calculated from the age specific mortality rates in the WHO data for 1966 (mid-point of the study). This information may be summarized by the Standardized Mortality Ratio (SMR) : SMR = Observed No. of Deaths Expected No. of Deaths SMR = --l8 - 0.77 = 77% 23.23 Since the incidence of leukemia i s low, and the study population is relatively small, chance factors may have produced an SMR in this study which i s quite different from what might have been found if the study population had been infinitely large. Therefore, before any conclusions can be drawn about the significance of the SMR, we must estimate the most likely values of the true SMR. This must be done in two steps: 1. Th.e Standard Error (SE) has to be calculated by the following formula: SE = SMR Expressed as % x 100 No. of Expected Deaths -SESMK= %f%= 18.2i1i% 23.23 2. A confidence interval (CI) must be calculated. This is an interval which we can say with 95% confidence includes the true SMR, and utilizes the following formula:12 CI = SMR 2 (1.96 x SEWR) CI = 77 + (1.96 x 18.21) = 113% -77 (l.% x 18.21) = 41% Since the interval within which we are 95% confident that the true S M R lies also includes the value 100, we cannot with confidence rule out the possibility that the true SMR i s in fact equal to 100. For -Tablo 7. Employm L Annuitant Yamlean and Caw d Lwkoaia. Mihh A/s Nor* Ew E n 0 AG. En0 Wmum * En0 Ihlklu 'En0 Nedmbnds N.V. Erso Petrdmrn CO. La. PO Sbndrrd SAF Eno (Switzerland) lbn-lan 13,439 74.419 16,606 38.092 17.807 143,554 -74.299 5.060 *d Lrhrir cam 0 3 0 0 0 8 -6 1 Tall 383t7Ct 18 'Doa nd include annuitants tlnduda 58.436 Annuitant ManYcPrs (15% d Tdal) this reason, instead of saying that our leukemia experience i s better than the general population we prefer simply to say that it is probably no worse than one would find in the same age span of the population as a whole. The next question that arises i s whether this "normal" mortality experience for the disease under suspicion (leukemia) could be unremarkable when the total European experience i s considered and still conceal an abnormal figure in a given location. If we look at Table 7 again we s e e that four of the affiliates have no cases, one has one case and another has !hree cases. Si inspection we can say that these six do not have an abnormal mortality experience. This leaves Esso Petroleum Company Ltd. with eight cases and Esso Standard S.A.F. with six to be evaluated. Table 9 summarizes the results of tedious calculations utilizing employeeannuitant person-years and the 1966 WHO Age Specific Mortality Rates for these two affiliates. This close approximation of observed to expected deaths is of particular interest in the case of Esso Petroleum. This affiliate has been keeping good followup records on annuitant deaths since 1964 (8 of the 10 years included in the study) which gives G :,oa;snable confidence that the observed number of - -Tabk 1. Er)rcttd n. Obmtved Deaths From Lwkenir in 1 Europoan Affilialn 1W 1971. h# 1) T i (Iw 1519 2424 2529 3.34 3539 w 4549 SOY 55.59 60.64 6169 10.14 7119 Lw84 a5 ~ n dOW 1 d LI m Edhtd @ r48sd TIhl ?maVur 3.248 15,592 36,707 49.051 47.102 51.000 48.726 37.681 38.049 25.960 11.519 8.064 5,610 3.214 -1.753 3u3n @ sW*' -m Rata(WH&l%6) 2.2 1.1 2.7 2.1 3.0 3.4 4.6 6.1 8.0 12.4 16.8 20.6 28.5 -37.1 30.0 Eqcld 0.07 0.26 0.99 1.03 1.41 1.73 224 2.30 3.04 3.22 1.94 1.66 1.60 -1.21 0.53 2323 Ohmd II.dDolhs 1 0 1 1 1 3 1 0 0 3 ')1( 4 3 (1) 0 0 0 7 18 Journal of Occupational Medicine/Vol. 16, No. 6lJune 1974 379 i i- Milm Esse Pel. Ca LM. EPO SM. SAF. ?anm T a r s 143.5% 74,299 ukemia deaths is correct. Our overall appraisal of the data for the eight individual affiliates is that there i s no abnormal concentration of leukemia in any one geographic location. Since we were unable to demonstrate any meaningful difference i n the leukemia mortality experience in one affiliate compared to the others, we decided to test the experience of the "exposed" versus the "non-exposed" fractions of the total population. Unfortunately we did not have the age distribution of these two groups for Europe as a whole to calculate "expected" vs. "observed" deaths in the same way we did for the total employeeannuitant population. A simple calculation can be made using the total age adjusted "expected" deaths (23.23) muitipiied by the fraction: "Exposed" person yearsiTotal person years and "Non-exposed" person yearslTotal person years with the results shown in Table 10. Although this indicates a slight preponderance of "observed" over "expected" deaths in the exposed group, the CI extends from 37 to 205, which translated in terms of expected deaths gives a range of 2.5 to 13.5 hence the observed difference i s not significant. One final attempt was made to determine whether there was a difference between the exposed and non-exposed categories. Esso Petroleum Company Table 10. - Expected vs Observed Leukemia Deaths in Exposed and LwEmpored 6rw~1 Expected NQ of Deaths 6.6 16 7 Dbxrved No of Deaths 8.0 S.M.R. 121. 60 From Leukemia in "Exposed" and "NonExpd" Portions of Euo P e t 13.Ltd. Population E r p d Ilon-trpard I Expected No. of Deaths Obrerved No. d Deaths SMR 4.62 5 108 4.33 69 IkdDrtk 8.94 5.56 Ikdhlb 8 6 Sm 09.5 107.7 which, as we have noted, probably has the most accurate figure for leukemia deaths has exposed and non-exposed populations which are approximately equal ("exposed"= 70,570 person years; "non-exposed" = 72,984 person years). We assumed that all person years i n the age groups 15 to 19 and 20 to 24 were "unexposed" since our criteria for ex- posure included at least 5 years in an "exposed" job. We then allocated the person years for the age groups 25 to 85 plus in the ratio 70,570 63.349::exposed: non-exposed, using the age distribution for the entire company. The calculations for the "expected" number of leukemia deaths and their SMR's were made as previously described. Table 11 shows the results of this calculation. Our conclusion is that we could not prove that the number of observed leukemia deaths exceeded the expected number in the exposed group at the 95% confidence level. Although we had no convincing evidence that the leukemia mortality in the 8 affiliates in the study was abnormal we attempted to relate the observed mortality rates to the potential benzene exposure in the group. Since we lacked information of the exposure levels measured by multiple sampling we approached this indirectly. In our original questionnaire we had requested data on the percentage of benzene in products processed or handled in various tvpes of operations - i.e. refining, transportation etc. Table 12 shows the form in which this data was reported by one affiliate: In a number oi the affiliates this information was much more varied. We attempted to reduce this complex mass of information to relatively simple terms by using a Benzene Index as shown in Table 13. Table 14 shows the Benzene Index and the crude mortality rate for leukemia in each affiliate. If this data is plotted with the Crude Mortality rate on the ordinate and the Benzene Index on the abscissa there i s no demonstrable relationship between the two (Figure 3 ) . One conclusion i s that low levels of potential benzene exposure are unrelated to leukemia mortality Other possible explanations are (1) The amount of benzene handled in our operations is not related to the actual exposure of employees (2) The criteria for estimating exposure are not uniform from one affiliate to another This emphasizes the point that more data are needed from our industrial hygienists to classify lob exposures to potentially hazardous materials Discussion and Conclusions Despite the problems of data collec- tion outlined in the early part of this presentation, the resultant necessity for making a number of assumptions and recourse to what some may consider "statistical manipulation", we beiieve that this effort was worth the time and energy on the part of all concerned. We believe that two oblectives were attained: 1. The demonstration that the occurrence of leukemia in a population of 38,000 working in a variety of petroleum and petrochemical operations, some of whom were exposed to low levels of benzene over a 10-year period was not abnormal compared to the general population in the countries concerned. 2. The demonstration of the need for improvement in the recording and storage in a readily accessible form of biological observations, job histories, occupational exposures and demographic data. In practical terms this means: A. The adoption of a form for recording the results of periodic examinations, the causes of sickness disability and death of employees which can be entered and stored in a computer system. In order for this information to be collected centrally where it can be of maximal value the information collection system should be uniform or at least compatible. B. The development of a method for acquiring information on the cause of death of annuitants. In affiliates where the company is paying all or part of an annuity there must be some mechanism whereby the Employee Relations Department is notified of the death of such individuals. If possible there should be a requirement that the company receive a death certificate. If so, a copy should be forwarded to the Medical Department where immediate follow-up can be made with the attending physician to 380 Epidemiological Survey of Leukemia and Potential Benzene Exposureflhorpe verify (and clarify) the diagnosis. This is particularly important where the possibility of a neoplasm exists, (as demonstrated in our present study by our limited information regarding the type of leukemia) Where the annuitant has no continuing contact with the company, strong efforts should be made to enlist the cooperation of public health authorities to obtain a periodic report of annuitant deaths. This should not be an impossible task if a social security number identity system is in effect. It is recognized that in some countries current laws forbid the transmittal of medical information to the company. In view ot the current interest in the epidemiology of occupational disease the reversal of bureaucratic attitudes may be a less formidable task than we have believed to date. Efforts in this direction should certainly be made. C. In a number of affiliates there has been a program to develop personnel data banks. Medical Directors and Advisors should consult w i t h their managements about the possibility of entering good job histories and potential hazardous exposures into this system. Observations of the industrial hygienist might well form part of the input into this data bank. D. Computer soft-wear programs should be developed to obtain yearly summaries of the distribution of the em- ployee and annuitant population in 5- year classes which correspond with those used for reporting census figures and WHO mortality data. While this concept of improving information storage and retrieval may appear unduly idealistic and expensive we are convinced that it i s essential and a swnd investment for the future. We are in the early stages of increased government and public health interest in regulating worker exposure to a multitude of environmental factors. In many instances regulations are proposed on highly theoretical considerations. We may expect that unrealistic regulations wilt tx passed i f there is no sound in- formation available in the formative stage of such laws or decrees. The chief source of such data with long term statistical validity will be industries or companies such as ours which have had good medical programs for years. The irony of the present situation i s that we probably have literally tons of data in /' / / /- / /- / / / / / / / / / / / e / / / a, a 100 aI 200 300 BENZENE INDEX 400 -F; 3. Benzene index vs. crude leukemia mortality rate. I@ 500 -1962 1-10 - 1%3 2-10 1%4 2-10 %la 1-Lo 1-10 1-10 l%S 2-10 5-100 1-10 1% 2-10 5-100 1-10 1%7 2-10 5-la 1-40 196% 2-10 S-1a 1-10 1969 2-10 5-100 1-10 Ism 2-10 5-10 1-10 1971 2-10 Cloo 1-10 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1-5 1 1 I mw Chuniuls TrmporWirn LrMin; AI A1 AI AI B=M B=AB B = AB 82 Journal of Occupational MedicineDol. 16, No. 6/June 1974 381 c c our records but no practical way to find or analyze it without the expenditure of thousands o i man-hours and the loss of time at critical periods. We believe this situation can and should be improved. This study would not have been possible without the wholehearredsupport of the Medical DirPcton of the affiliatesinvolved Spectal thanks dre due to k.46 )udith Yadelmann M A , &ostatistician of the E x x m Corporation Medical Department who providedinvaluable guidance in the area of statistical theory as well as devocing many hours to the calculations involved 1. Browning E: Toxicity and metabolismof industrial solvents. Amsterdam, Elsevier, 1965. 2. Damedhek W, Gunz F Leukemia. New York, Crune & Stratton, 1959. 3. Vigliani EC, Saita G: Benzene and leukemia. N Engl / Med 271;872-876, 1964. 4. Cavignaux L: Les intoxications confirmees. Cah Med lnterprof 228-31, 1962. 5. Aksoy M, et al: Acute leukemia due to chronic exposure to benzene. Am / Med 52:160-166, 1972. 6. Aksoy M, et al: Haematological effects of References chronic benzene poisoning in 217 workers. Brit / Ind Med 28:296-302, 1971. 7. lshimaru T, et al: Occupational factors in the epidemiolog'y of leukemia in Hiroshima and Nagasaki. Am / Epidemrol 93:157-165. 1971. 8. Cronkite EP: Evidence for radiation and chemicals as leukemogenic agents. Arch Environ Health 3:297-303. 1961. 9. Sherwwd RI: Occupational hygiene in aromatics plants. Ann Occup Hyg 14:125-135. 1971. io. Sherwood RI: Evaluation of exposure to benzene vapour during the loading of petrol. Brir / lnd Med 29:65. 1972. 11. Parkinson CS: Benzene in motor gasoline - an investigation into possiblehealth hazards in and around filling stations and in normal transport operations. Ann Occup Hyg 14:145-153, 1971. 12. W. Caffey: Personal Communication who derived it from a paper by CL Chiang: Standard Error of the Age - Adjusted Death Rate Vital Statistics Special Report 47.275-285, 1961.