Document OEb7ORB0RLV3Ljp00Mm38dG9p

Occupation and the High Risk of Lung Cancer in Northeast Florida WILLIAM J. BLOT, PhD,* JOHN E. DAVIES, MD,t LINDA MORRIS BROWN, MPH/ CHARLES W. NORDWALL.t EVA BUIATTI. MD.* ALAN NG, MD,t AND JOSEPH F. FRAUMENI, Jr.. MD* A case-control study Imohing interviews with 321 male patients with lung cancer and 434 controls, or their next of kin, was undertaken to identify reasons for the high lung cancer mortality along the northeast coast of Florida. In Duval county (Jacksonville), the age-adjusted rate for lung cancer, 1970 1975, among white males was the highest of all urban counties in the United States. Increased risks on the order of 40-50% were associated w ith employment in the shipbuilding, construction, and lumber/ wood industries, particularly among workers with reported exposures to asbestos or wood dust. Excess risks were also linked to fishing and forestry occupations, although the numbers of cases involved were small. Occupational factors did not appear to fully account for the area-wide excess of lung cancer, but no evidence was found to implicate smoking habits, migration patterns, or diagnostic and reporting practices as factors responsible for the exceptional mortality rates. Cancer 50:364-371, 1982. county-by-county survey of lung cancer mor counties along the northeast Florida coast (Flagler, St, A tality in the United States during the period Johns, and Nassau), and the remaining 3052 counties 1950-1969 revealed a clustering of elevated rates alongof the contiguous United Slates, using data from the the southeast Atlantic seaboard, primarily among white National Center for Health Statistics and the Bureau males, with Duval county (Jacksonville) in Florida hav of the Census. Not all deaths for 1972 were recorded ing one of the highest rates in the United States.'*3 at the county level and therefore are not included in the Case-control studies in coastal Georgia and Virginia calculations. Methods of calculating age-adjusted rates have linked the excess risks of lung cancer in those areas are described elsewhere.1 to employment in shipyards, particularly those oper The case-control study consisted of two components. ating during World War II.4-5 To identify the factors The first was a hospital-based study in which all male responsible for the exceptionally high rales of lung can residents of Duval county new ly diagnosed with primary cer in Duval county, a case-control study involving in lung cancer at one of 13 Jacksonville hospitals during terviews with lung cancer patients or their next of kin the period April 10, 1978 through April 9, 1979 were was carried out. prospectively ascertained. Tissue slides were sought for Methods each case and independently reviewed by one of the authors (A.N.). Two controls matched by age (2 To update lung cancer mortality statistics reported years), race, and hospital were selected for each case for the period I95G-1969,3 mortality rates were com from closest (in time) admissions of male county resi puted for the period 1970-1975 by sex, race (white, dents to those same hospitals for diagnoses other than nonwhite) and age for Duval county, its neighboring lung cancer, chronic respiratory disease, or psychiatric conditions. The second component of the study used Environmental Epidemiology Branch, National Cancer Institute, Bcthesda. Maryland. t Departments of Epidemiology and Public Health, and Pathology, University of Miami School of Medicine, Miami. Florida. Address for reprints: William J. Blot, PhD, Head, Analytical Stud ies Section. Environmental Epidemiology Branch, Landow Building, Room 3C07, Bethesda. MD 20205. The authors thank (he study members, their physicians, and the staffs of the participating hospitals; Dr. P. C. Cowdery of the Duval County Health Department for advice and support; Ms. M. Malone, and Ms. A. Toledo of Westat Research. Inc., for field operations: Dr. B. J. Stone. Ms. L. Kammerman. and Mr. T. E. Aldrich for computer assistance: and Ms. T. McKinney for manuscript preparation. Accepted for publication May 9, 1981. mortality files for 1976 at the Slate Department of Health and Rehabilitative Services to identify cases and controls. The cases were all deaths attributed to primary lung cancer among male residents of Duval, Flagler, St. Johns, and Nassau counties, while an equal number of controls, matched by race, age, and county of resi dence, were randomly selected from deaths attributed to other causes (excluding lung cancer or chronic res piratory disease). ' Identifying information from the death certificates was used to locate next of kin of cases and controls in O0G8-543X/82/O7J5/0364 S0.90 <5> American Cancer Society 364 Ho. 2 Occupation and Risk of Lung Cancer Blot a ol. 36 5 V > it. ies the mortality series for interview. Interviews with pa tients in the hospital series were sought in the hospital, vkhere feasible, or in the patients' homes. Interviews were conducted by local interviewers under the super vision of a professional survey organization, with inter viewers not informed of the case-control status of the study subject. The questionnaire sought information on tobacco consumption, and residential, medical and oc cupational histories. The name of the employer, job ti tle, and duties performed were asked for each job held for six months or longer. The occupational section of the interview included a checklist of certain industries and materials handled. The industrial data were cate gorized using a scheme developed for this study based on Standard Industrial Classification codes.6 Respon dents were also asked about antecedent lung diseases (including chronic bronchitis, emphysema, recurrent pneumonia, asthma, pleurisy, and tuberculosis). The measure of association between variables of in terest and lung cancer was the relative risk (RR), ap proximated by the odds ratio.7 The data from the hos pital and mortality series were analyzed separately, but summary RR combining the results were obtained by the Mantel-Haenszel method.8 Results ied During 1970-1975, Duval county had the highest .he lung cancer mortality rate among white males of any tes metropolitan county in the nation (Table 1). The annual age-adjusted death rate of 93.2/105 far exceeded the its. | national average of 58.6/105. As shown in Figure 1, ale ` rates in Duval county were higher than the United ary ! States rates at all ages. The age-adjusted rates among ing white males in the smaller neighboring counties of Flag ere ler, St. Johns, and Nassau were also high, being 100.9, for 72.1, and 103.0, respectively. Mortality from lung can the 2 1 i cer among nonwhite males in Duval county was less than among whites, but the rate of 77.7 was in excess ase . . i of the national rate for nonwhite males of 67.4. SSI- During the one-year period between April 10, 1978 han 1 and April 9, 1979, 181 incident cases of primary lung trie cancer and 342 controls* among male residents were sed identified from admissions to the 13 Jacksonville hos of I pitals. Review of the mortality register for 1976 re and vealed 217 deaths attributed to primary lung cancer in .ary . the four-county area, 89% from Duval county, while an :ler, equal number of deaths from other causes served as ber controls. The controls in the hospital series represented esi- admissions for other cancers (3%), cardiovascular disjte Tai*ll I. Listing of the Ten Metropolitan Counties* in the United States with the Highest Age-adjusted Rates of Lung Cancer Mortality Among White Males. 1970-1975+ County Mortality rate (deaths/yr/IO') Duval. FL St. Louis City, MO Baltimore City, MD Chesapeake.t VA Orleans. LA Mobile. AL Jefferson, KY James Cily, VA Chesterfield,1 VA Marion, IN 93.2 90.9 88.4 87.2 86.1 83.8 82.8 80.4 79.3 77.6 Includes all counties with at least 500.000 person-years of ob servation among white males during 1970-1975. t Deaths for 1972 are excluded since not all were ascertained for this year. . $ Includes the independent cities of Norfolk and Portsmouth. Includes the independent city of Newport News. 1 Includes the independent city of Richmond. eases (37%), digestive diseases (19%), accidents (6%) and other conditions (35%). The controls in the mor tality series represented deaths due to other cancers (11%), cardiovascular diseases (67%), digestive diseases (8%), accidents (3%), and other causes (11%). A total of 807 interviews were completed (4% of the V * Twenty controls were not assigned, eight from one hospital due ates to inaccessibility of discharge records, 12 corresponding to six cases Fig. I. Lung cancer mortality rates 1970-1975. by age among white s in who decided not to participate in the study. males in Duval county and the United States. f 366 Canclr July 15 1982 V'ol. 50 Taiii t 2. Numbers of Cases and Controls from the Hospital and Mortality Series According to Age, Race. Education Level, and Slate of Uirth Mortality series Hospital scries Indicator Age <50 50-59 60-69 70+ Race White Biack Education <12 12 + State of birth Florida Other southern state Elsewhere Cases No. % 18 10.6 34 20.1 64 37.9 53 31.4 132 78.1 37 21.9 93 55.0 76 45.0 44 26.0 90 53.3 35 20.7 Controls No. % 20 12.7 30 38.0 60 38.0 48 30.3 in 74.1 41 25.9 80 50.6 78 49.4 39 24.7 80 50.6 39 24.7 Cases No. % 14 9.2 49 32.2 53 34.9 36 23.7 121 79.6 31 20.4 98 64.5 54 35.5 47 30.9 67 44.1 38 25.0 Controls No. % 27 9.8 77 27.9 100 36.2 72 26.1 217 78.6 59 21.4 154 55.8 122 44.2 72 26.1 133 48.2 71 25.7 target sample) with the response rate somewhat higher for cases (86%) than controls (83%). In the hospital series the next-of-kin provided a higher percentage of interviews for cases (42%) than controls (13%), because lung cancer is often rapidly debilitating and fatal. For this reason, analyses stratified the hospital series with respect to the type of respondent (self versus next-ofkin) unless such division resulted in exceptionally small sample sizes. The next-of-kin interviews in both series were most often (71% for both cases and controls) con ducted with the wives of the study subjects. Of the total number interviewed, ten were excluded because of un satisfactory completion of the questionnaires. An ad ditional 42 who were short-term, i.e., less than five-year residents in northeastern Florida were also deleted, bringing the final study group to 755 subjects, 321 cases and 434 controls. Table 2 shows the distribution of cases and controls according to age, race, educational level, and state of birth. Subjects in the hospital series were younger (median, age 63 years) than subjects in the mortality series (median, age 65 years), but cases and controls Table 3. Relative Risk (RR) of Lung Cancer for Ever Employment in 15 Industrial Categories According to Source of Ascertainment and Type of Respondent Source of ascertainment (respondent type) Mortality Hospital (next of kin) Hospital (self) Summary Industry Case Control RR* Case Control RR* Case Control RR* RR Agriculture 51 34 1.6 22 Forestry 4 3 1.3 0 Fishing 5 1 4.8 0 Construction 58 48 1.2 16 Lumber/ood 14 10 1.3 7 Metal 22 18 1.2 15 Shipbuilding 35 25 1.4 15 Paper/pulp 5 10 0.5 4 Chemical, misc. mfg. 31 35 0.8 11 Transportation 77 67 1.1 36 Wholesale/retail 58 74 0.6 19 Personal service 21 27 0.7 n Professional 82 68 1.3 28 NEC 17 11 1.5 3 Military . 65 54 1.2 24 * All risks are relative to those never employed in the industry, t P < 0.05 (one-sided). 9 1.6 27 1 --5 1 --4 9 1.0 35 2 2.1 14 4 2.6 14 5 2.0 21 05 10 0.6 3! 18 1.4 49 18 0.4 41 8 0.8 13 12 1.6 44 5 0.3 7 12 1.3 40 % P < 0.01 (one-sided). 78 0.9 1 14.3 2 5.6 63 1.8 26 1.5 38 1.0 44 1.4 12 1.1 80 1.1 129 1.1 138 0.6 43 0.8 122 1.0 12 1.6 no 1.0 1.3 2.3t 3.2t !.4t 1.5 1.2 1.5* 0.9 0.9 1.1 0.6$ 0.7 1.2 1.2 1.1 I. 50 -- > . 4 No. 2 Occupation and Risk of Luncj Canclr Blot cl al. J67 Tahi.l 4. Relative Risk (RR) of Lung Cancer Associated with Employment in Selected Industries According to Cigarette Smoking Category Smoking category Industry Shipbuilding Construction Lumber/wood 1 Wholesale/rctail Ever employed No Yes No Yes No Yes No Yes Non. <`'5 pk/day ~ Case Control RR* 17- no 1.0 5 18 2.7 14 97 1.0 8 31 1.5 21 115 1.0 l 13 0.6 17 62 1.0 5 66 0.3 Case 117 33 101 49 130 20 99 51 '4-1 pk/day Control 132 37 112 57 155 14 75 94 RR* 6.9 7.2 6.2 6.1 5.1 10.6 5.4 2.6 . >1 pk/day Case Control 91 92 32 17 77 83 46 26 110 100 13 9 69 50 54 59 RR* 7.7 14.4 6.3 13.5 6.5 14.7 5.7 4.1 All risks are adjusted for source of ascertainment and are relative nonsmokers and smokers of less than one-half pack (pk) per day and to those for men who were in the smoking category consisting of who were never employed in the particular industry. J 2 7 f unn ad -year. 'ti ntrols ate of unger rtality mtrols :nment in each series were similar in age and race because of the matching employed in the design. The cases tended to have fewer years of formal education than the con trols, and were somewhat more often born in the state of Florida. However, differences according to duration of residence in the Jacksonville area were small and not statistically significant. Table 3 lists the RR according to ever employment in each of 15 industrial categories. Among the major industries in the area, 40-50% elevations in risk were associated with shipbuilding, construction, and lumber/ wood employment, while decreases in risk were found for wholesale/retail trades. There was considerable overlap among the former three industries. About 40% of the men ever employed in the shipbuilding or the lumber/wood industries also worked at some time in construction. Elevated risks, however, were seen among those with no construction employment (RR = 1.5 for shipbuilding, RR = 1.6 for lumber/wood), while an increased risk associated with construction work (RR = 1.3) was found among those with no history of ship building or woodworking jobs. RR of 3.2 and 2.3 were found for ever employment in the fishing and forestry industries, respectively, but the numbers of individuals involved were small. The RR for usual employment in the 15 industrial groups, i.e., the industry where an individual worked the longest number of years, showed similar patterns. Table 4 shows the RR for selected industries ac cording to levels of cigarette smoking. The higher risks associated with shipbuilding were seen in each smoking category, but the largest absolute excess risk was among heavy smokers. The summary RR adjusted for smoking and source/respondent status was 1.4 (one-sided P Timary RR U 2.3+ 3.2+ 1.4+ 1.5 1.2 1.5+ 0.9 0.9 U 0.6$ O' 1.. Table 5. Relative Risk (RR) of Lung Cancer by Number of Years Worked in Shipbuilding, Construction and Lumber/Wood Industries According to Source of Ascertainment Source of ascertainment industry ' Years employed Case Mortality series Control RR Case Hospital series Control RR Summary RR Shipbuilding Construction lumber/wood Never 134 133 1.0 116 227 1.0 1-9 24 21 1.1 26 39 1.3 10+ 8 4 2.0 8 9 1.7 UK 3 0-- 2 1-- Never 111 110 1.0 101 204 1.0 1-9 24 . 19 1.3 28 39 1.5 10+ 32 29 1.1 21 33 1.3 UK 2 0-- 2 0-- Never 155 148 1.0 131 248 1.0 1-9 11 7 1.5 14 22 1.2 10+ 2 2 1.0 6 6 1.9 UK 1 1-- 1 0-- 1.0 1.2 1.8 -- 1.0 1.4 1.2 -- 1.0 1.3 1.6 -- UK: unknown duration. 368 Cancer July 15 1982 vol. 50 Taih ! 6. Relative Risk (RR) of Lung Cancer Associated with Employment in the Shipbuilding and Construction Industries According Reported Exposure to Asbestos and Type of Respondent * ' Industry Shipbuilding Construction Type of respondent Ever Asbestos Next-of-kin Self Sumnia employed exposure Case Control RR Case Control RR RR No Yes No Yes _ No Yes _ No Yes 183 165 1.0 67 39 27 1.3 12 11 3 3.3 9 159 138 1.0 53 59 53 1.0 24 15 4 3.3 11 195 1.0 30 1.2 14 1.9 176 1.0 45 9.8 18 2.0 10 L3 2.3 1o 12 2.5 = 0.06), with 90% confidence limits of 0.99-1.90. The RR for construction and lumber/wood employment were not elevated in every smoking category, but the decreases associated with wholesale/retail trades were consistently observed. The summary RR and corre sponding one-sided ^-values and confidence limits were 1.4 {P = 0.04; 1.03-1.8) for construction, 1.7 (P = 0.04; 1.04-2.7) for lumber/wood, and 0.5 (P < 0.001; 0.40 0.70) for wholesale/retail. The percentage of smokers among the small number of the men employed in fores try or fishing occupations was not elevated. Most of the men employed in shipbuilding worked in the industry for limited periods, often during World War II. Only 20% were employed for at least ten years, but among these the RR reached 1.8 (Table 5). The RR associated with lumber/wood jobs was also some what higher among the minority who worked ten or more years, while no trend according to duration of employment was detected for the construction industry. The RR for these industries were also calculated ac cording to age, race, education level, and duration of residence in northeast Florida. There were no consistent trends in the RR with age. The RR for shipbuilding were higher in blacks than whites, among those with less than a high school education, and among long-term (35+ years) residents. The elevated RR for construction jobs were limited to those with less than 12 years ed- ucation. . In the checklist of materials handled at work, the RR was about two for persons exposed to asbestos or wood dust. The relative excess risk was independent of source of ascertainment, although self-respondents reported considerably higher exposure rates than next-of-kin; among the controls, 15% of self-respondents versus 4% of next-of-kin reported asbestos exposures, with corre sponding percentages of 18% versus 10% for wood dust. As shown in Table 6, asbestos exposures accounted for much of the higher RR associated with shipbuilding and construction. The excess for lumber/wood jobs was primarily among workers with wood dust exposures (RR 1.9) than among those without (RR = 1.2), Table 7 shows the RR of different histologic types of lung cancer for the selected industrial groups and exposures using data from the hospital series (based on classifications of Jacksonville pathologists). The slide review revealed that all cases were consistent with the diagnosis of lung cancer. There was agreement as to histologic type for 78% of the cases, with our review indicating a smaller percentage of squamous cell and unspecified carcinomas, and a larger percentage of ad enocarcinoma and anaplastic small cell and large cell carcinomas. The increased risks for construction, ship- ! | \ I | I j * ! { ` Table 7. Relative Risk (RR) for Histologic Types of Lung Cancer Associated with Employment in the Construction. Shipbuilding, and Lumber/Wood Industries, and Reported Exposures to Wood Dust and Asbestos from the Hospital Series Indicator Others Small cell Adeno and Controls Squamous RR anaplastic RR carcinoma RR unspecified RR Industry Construction Shipbuilding Lumber/wood Exposures Wood dust Asbestos No Yes No Yes No Yes No Yes No Yes 202 71 225 48 246 27 226 47 236 37 36 1.0 15 1.0 24 1.0 26 15 1.2 10 1.9 10 1.8 16 38 1.0 19 1.0 26 1.0 33 13 1.6 6 1.5 8 1.4 9 42 1.0 22 1.0 30 1.0 37 9 2.0 3 1.2 4 1.2 5 1.0 1.9 1.0 1.2 1.0 1.2 41 1.0 15 1.0 28 1.0 32 1.0 10 1.2 10 3.4 6 1.1 10 1.5 44 1.0 17 1.0 26 1.0 36 1.0 7 1.0 8 3.0 8 2-.0 6 1.1 50 - i \1 I iiary ,j --- \ $1 j *> \ 5i ^ Occupation and Risk oi Lung Canclr Blot ct al. 369 X-\BI l 8. Relative Risk (RR) of Lung Cancer Associated with Pritir History of Chronic Lung Disease According to Type of Respondent " Type of respondent Next-of-kin Self Antecedent lung disease Case Number reporting disease Control RR RR* Case Number reporting disease Control RR RR* Empyema Chronic bronchitis pieurisv Asthma 70 26 2.8 2.1 19 31 1.9 1.4 28 16 1.5 1.3 8 22 1.0 0.7 23 13 1.5 1.2 10 21 1.3 1.2 18 6 2.6 2.8 3 22 0.4 0.3 Relative risk of lung cancer associated with prior occurrence of the particular lung disease adjusted for cigarette smoking category. ed- j building, and Iumber/wood occupations did not seem specific for particular cell types. However, there were RR ood jrce ned -kin; ;4% Dire dust. dfpr^v, | higher risks for small cell carcinoma following expo sures to wood dust (RR = 3.2) and asbestos (RR = 3.0). In response to questions on chronic respiratory ill . nesses, 32% of the controls (in both the next-of-kin and I self interviews) responded positively, compared to 53% and 47% of the cases, respectively. Emphysema was more closely associated with lung cancer than were the other lung diseases most often listed, i.e., chronic bron chitis, pleurisy, and asthma (Table 8). The RR for em bures 2). types ; and ed on slide h the as to eview 1 and yf ad;e cell . ship- physema were as high or higher for both self and nextof-kin respondents who reported emphysema prior to 1970, i.e., at least six years prior to the onset of cancer, than after 1970. The magnitude of the association with emphysema was reduced when cigarette smoking was taken into account, but the positive link still remained. The smoking and respondent-adjusted RR of lung can cer associated with antecedent emphysema was 1.9 (one-sided P < 0.001 with 90% confidence limits, 1.5 2.4). The reporting of emphysema was higher among both cases and controls who had worked in construction or shipbuilding jobs, but the elevated RR of lung cancer associated with these industries were observed among men with and without reported emphysema. g, and RR 1.0 1.9 1.0 1.0 1.5 1.0 i.i Discussion The patterns of cancer mortality by county during the period 1950-1969 revealed a cluster of elevated rates for lung cancer along the southeast Atlantic coast.'*3 Our updated statistics for northeastern Florida counties show that the excess in this region has become more pronounced in 1970s. Indeed, lung cancer mor tality among white males in Duval county in the period 1970-1975 was the highest recorded among all met ropolitan counties of the United States, and was greater than the national average by more than 50%. Mortality was also elevated among blacks, but not to as great an extent. The elevated risks associated with employment in shipbuilding, construction, and wood-related industries suggest that occupational exposures account for at least part of the high lung cancer mortality among male res idents of northeast Florida. The RR for these industries are not high enough, however, to fully account for the area-wide excess among white males, but no other fac tors could be implicated in this study. The percentage of controls who smoked cigarettes was somewhat higher than the national average,9 but the controls included persons hospitalized or deceased from conditions known to be related to smoking. The authors recently com pleted a population-based telephone survey in Georgia indicating no difference in the percent of cigarette smokers (ever or current) among coastal versus inland white male residents aged 45 and older, and the percent of smokers in the Jacksonville controls resembled the Georgia statewide average. The excess lung cancer in northeast Florida does not seem related to the influx of residents from outside the area, since no consistent gradients were observed with duration of residence; risks were in fact slightly higher among those born in the state than elsewhere. It is difficult to evaluate local variations in diagnostic and reporting practices, but the clustering of elevated mortality from lung cancer ex tends beyond Jacksonville to involve most of the south east Atlantic coast, with a gradual reduction in rates towards the inland counties. Furthermore, an indepen dent pathologic review uncovered no misdiagnoses of lung cancer, so that overreporting was not a problem in this area. General environmental factors, e.g., sulfate air pollution, excessive humidity, and radon exposures in homes and buildings may be involved, but additional study using a different approach would be required to evaluate such potential risk factors. The relation of shipbuilding to lung cancer has now been observed in three areas along the southeast At lantic coast: Georgia,4 the Norfolk-Newport News re gion of Virginia,5 and the Jacksonville area of Florida. Interview studies in the first two areas found increased risks of lung cancer, on the order of 60-70% after taking smoking into account, associated with employment in the shipbuilding industry during World War II. In all three instances a sizable percentage (20-35%) of the 370 Cancer July 15 1982 Vol. so [ No- cases were employed in area shipyards, often for only limited periods during the 1940s. In Jacksonville em ployment reached a war-time peak of about 20,000, then dropped to less than 2000 in the 1950s and there after.10-11 The elevated risk associated with shipbuilding in this study was mainly among those reportedly exposed to asbestos. Some excess was found among those without exposures reported, but the differences in direct versus next-of-kin interviews suggest that knowledge of ex posure by the next of kin may be incomplete. The work ers themselves may also have been unaware of asbestos exposure. Only a minority of those employed in ship building were said to have worked with asbestos ma terials, but asbestos was extensively used in insulation and fireproofing in large ships and inadvertant exposure appears to have been widespread.12 In northeast Florida, as well as coastal Georgia and Virginia, the excess risk of lung cancer among shipyard workers was greatest among heavy cigarette smokers, consistent with the syngeristic interaction described for asbestos exposure and smoking.13 Only a few cases of asbestosis or mesothelioma were found through pathology and diagnostic index reviews at the participating hospitals in the Jacksonville area. This resembles the situation in Georgia, in contrast to the 15-fold excess of mesothelioma among shipyard workers in Virginia.14 It is not clear whether this re gional variation in mesothelioma reflects diagnostic practices or differences in asbestos exposure, e.g., in tensity or type of fiber. Each region was involved in the construction of large ships (mainly liberty cargo vessels in Florida and Georgia and combatant naval ships in Virginia) requiring great amounts of asbestos materials. The shipbuilding industries in Florida and Georgia, however, began operations during World War II, while the industry was established in coastal Virginia prior to World War I.15 Since asbestos-induced mesothelioma is.characterized by latent periods of 35 years and more, the increase is perhaps yet to be seen among former workers in Florida and Georgia. In contrast, the shorter latent periods for asbestos-induced lung cancer16 have already permitted recognition of an excess in all three states. The small elevation in RR associated with construc tion may also be related to asbestos exposure, which was reported by about 30% of the construction workers who were directly interviewed. A death certificate sur vey in coastal Georgia found construction mentioned as the usual industry about twice as often for deaths attributed to lung cancer," although some of those men also had prior shipyard employment.4 In our study, the excess risk associated with construction work persisted after adjusting for smoking habits and employment in shipyards. The findings are consistent with previous case-control studies18*19 and occupational mortality sur- veys,20'24 although limited in nature, which have sug gested an increased risk of lung cancer in construction workers. Asbestos has been a suspected agent,25 al though polycyclic hydrocarbons appear involved in some categories of work such as roofers.26 The small elevation in risk for workers in the lumber/ wood industry (primarily sawmill laborers) was mainly among those with wood dust exposure, reported by about two-thirds of those directly interviewed. Nasal cancer is a well-established disease among furniture makers, especially those with exposure to hard wood dusts,27 and a relationship to laryngeal cancer has also been reported.28 This industry was recently implicated in a small case-control study of lung cancer in Swe den,29 but excess lung cancer has not generally been reported among furniture makers, sawmill workers, lumbermen, or others handling wood products.30 Among death certificates for lung cancer in coastal Georgia, there was an excess frequency of wood-related occu pations, principally in rural areas.17 These occupations also included forestry jobs, a separate category in our survey where risk was also significantly elevated. The excess risk associated with lumber/wood employment was limited to smokers, raising the possibility of a cocarcinogenic effect. The increased risk for the occupation of fishing, al though based on small numbers, has been reported in other surveys.21*22 Reasons for this association are un known. It is also noteworthy that lung cancer was not related to work in the paper and pulp industry. This confirms the findings in coastal Georgia4 and suggests that the excess mortality from lung cancer reported in countries with a concentration of paper or pulp mills2 is unrelated to employment in this industry. A decreased risk of lung cancer was observed among men who held jobs in wholesale or retail sales. The deficit was not accounted for by lower smoking levels, but whether it is related to the "cleaner" work envi ronment or to other factors is not clear. Under the as sumption that workers in clerical jobs (including many in the sales industry) are unlikely to be exposed to occupational carcinogens, one case-control survey of hospitalized patients in Buffalo estimated the risks of cancer for various occupations relative to the risks for clerical workers.31 This procedure, not done in our survey, would have increased the magnitude of the RR for each of the high-risk industries. Although it is difficult to disentangle the effects of smoking, chronic obstructive lung disease has been reported to predispose to lung cancer.32 The connection between emphysema and lung cancer observed in our study was only partly accounted for by the relation of both diseases to cigarette smoking. Emphysema was more often reported for cases and controls who had worked in the shipbuilding and construction industries. Thus, although this study was not designed to identify \ * ^ | I i i | 1 I | \ | , I , I risk ther mon stru< port pred emp the r that have lung D edge aboi ably Vali' prox defir occu stud; Df stud; vious indu; parti the s assoc dusti rates that stud) tion} tribu 1. M ttlity fc Washin 2. Bl Indust- 3. h 1969. emmet 4. B emplo\ 299:62 5. B cers in Cancer 6. C ficaiior Office, 7. F York,' 8. h data fr 22:719 9. L' 1975. , 10. 1 and cot US De SO ; t )n .1 in r/ !y ^>y al re od .so ed ve:en TS, ng :ia, cuDns our The ent co- unnot This zests d in sills* long The vels, envie asnany o oc* hos' can* s for r surR for :t$ of en rset. a-' a was o had .striesientify No-2 Occupation and Risk or Lung Cancer Blot ct al. 371 risk factors for emphysema, nor to verify its occurrence, there is the possibility of occupational components com mon to both diseases. It is of interest that chronic ob structive lung disease and lung cancer have been re ported to cluster in families,35 suggesting a familial predisposition to both diseases. Since mortality from emphysema is relatively high among white males along the northeast Florida and Georgia coasts,54 it is possible that chronic pulmonary disease and its determinants have contributed to the area's excess cancer rates for lung cancer. Difficulty in recall by the patients and lack of knowl edge by their next of kin limited the amount of detail about occupational and other exposures, although prob ably not differentially between the cases and controls. Validation studies in other populations suggest that proxy respondents can provide adequate data on broadly defined variables, including smoking category and usual occupation,4-55,56 and the RR estimates in the present study were similar for self and next-of-kin respondents. Despite methodologic limitations, this case-control study in the Jacksonville area of Florida extends pre vious studies indicating that work in the shipbuilding industry, particularly during World War II, is at least partly responsible for the clustering of lung cancer along the southeast Atlantic coast. Increased risks were also associated with the construction and lumber/wood in dustries, which may contribute to the exceptionally high rates of lung cancer in northeast Florida. It seems likely that nonoccupational factors are also involved, but this study yielded no evidence that smoking habits, migra tion patterns, or diagnostic and reporting practices con tributed to the area-wide excess of lung cancer. REFERENCES I. Mason TJ, McKay FW, Hoover R, et al. Atlas of cancer mor tality for US counties: 1950-1969. DHEW Publ No (NIH) 75-780. Washington, DC, US Government Printing OfFice, 1975. 1 Blot WJ, Fraumeni JF Jr. Geographic patterns of lung cancer: Industrial correlations. Am J Epidemiol 1978; 103:539-550. 3. Mason TJ, McKay FW.U.S. cancer mortality by county: 1950 1969. DHEW Pub! No.(NIH) 74-615. Washington, DC, US Gov ernment Printing Office, 1974. 4. Blot WJ, Harrington JM, Toledo A, ei al. Lung cancer after employment in shipyards during World War 11. N Engl J Med 1978; 299:620-624. 5. Blot WJ, Morris LE, Stroube R, et al. Lung and laryngeal can cers in relation to shipyard employment in coastal Virginia. J Natl Cancer Inst 1980: 65:571-575. 6. Office of Management and Budget. Standard industrial classi fication manual 1972. Washington, DC, US Government Printing Office, 1972. 7. Fleiss JL. Statistical Methods for Rates and Proportions. New York, Wiley, 1973; 43-62. 8. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959; 22:719-748. 9. US Dept Health, Education and Welfare. Adult Use of Tobacco: 1975. Atlanta, Center for Disease Control, 1976. 10. Bureau of Labor Statistics. Wartime employment, production, *od conditions of work in shipyards. Bull No 824. Washington DC, US Department of Labor, 1945. 11. US Dept of Commerce. US census of population 1950 (1970). Washington DC. US Government Printing Office, 1952 (1971). 12. SelikofT IJ. Nicholson WJ, Lilis R. Radiologic evidence of as bestos disease among ship repair workers. J Ind Med 1980; 1:9-22. 13. SelikofT IJ. Cancer risk of asbestos exposure. In Hiatt HH, Watson JD, Winsten JA, eds. Orpins of Human Cancer. Cold Spring Harbor, New York, Cold Spring Harbor Laboratory, 1977; 1765 1784. 14. Tagnon I, Blot WJ, Stroube R, et al. Mesothelioma associated with the shipbuilding industry in coastal Virginia. Cancer Res 1980; 40:3875-3879. 15. Fasset, FG Jr. The shipbuilding business in the United States of America, vol 1. New York, Society of Naval Architects and Marine engineers, 1948. 16. SelikofT IJ, Hammond EC, Seidman H. Mortality experience of insulation workers in the Unites States and Canada. 1943-1976. Ann NY Acad Sci 1979; 330:91-116. 17. Harrington JM, Blot WJ, Hoover RN, et al. Lung cancer in coastal Georgia: A death certificate analysis of occupation. J Natl Cancer Inst 1978; 60:295-298. 18. Breslow L, Hoagiin L, Rasmussen G. et al. Occupations and cigarette smoking as factors in lung cancer. Am J Pub Health 1954; 44:171-181. 19. Levin ML, Kraus AS, Goldberg ID, et al. Problems in the study of occupation and smoking in relation to lung cancer. Cancer 1955; 8:932-936. 20. Office of Population Censuses and Surveys. Occupational mor tality 1970-1972. London, H. M. Stationery Office, 1978. 21. Milham S Jr. Occupational mortality in Washington State )950-71. DHEW Publ No (NIOSH) 76-175. Washington DC, US Government Printing Office, 1976. 22. Gottlieb MS, Pickle LW, Blot WJ, et al. Lung cancer in Lou isiana: Death certificate analysis. J Natl Cancer Inst 1979; 63:1131J137. 23. Milham S Jr. Mortality experience of the AFL-CJO Union Brotherhood of Carpenters and Joiners of America. DHEW Publ No (NIOSH) 74-129. Washington DC, US Government Printing Office, 1974. 24. DeCoufle P, Lloyd JW, Salvin LG. Causes of death among construction machinery operators. J Occup Med 1977; 19:123-128. 25. Englund A. Based on available data, can we project an ac ceptable standard for industrial uses of asbestos: No. Ann NY Acad Sci 1979; 330:219-222. 26. Hammond EC, SelikofT IJ, Lawther PL, et al. Inhalation of benzpyrene and cancer in man . Ann NY Acad Sci 1976; 271:116-- 124. 27. Acheson ED. Nasal cancer in the furniture and boot and shoe manufacturing industries. Prev Med 1976; 5:295-315. 28. Wynder EL, Covey LS. Mabuchi K. et al. Environmental fac tors in cancer of the larynx. Cancer 1976; 38: 1591-1601. 29. Esping B. Axelson D. A pilot study of respiratory and digestive tract cancer among woodworkers. Scand J Work Environ Hhh 1980; 6:201-205. 30. 1ARC monographs on the evaluation of the carcinogenic risk ofchemicals to humans. Wood, leather and some associated industries, Vol. 25. Lyon, International Agency for Research on Cancer, 1981. 31. Bross JD, Viadana E, Houten L. Occupational cancer in men exposed to dust and other environmental hazards. Arch Environ Hhh 1978; 32:300-307. 32. Rimington J. Smoking, chronic bronchitis, and lung cancer. Sr MedJl97U 2:373-375. 33. Cohen BH. Graves CG, Levy DA, et al. A common familial component in lung cancer and chronic obstructive pulmonary disease. Lancet 1977; 2;523-526. 34. Mason TJ, Fraumeni JF Jr, Hoover RN, et al. Atlas of mor tality from selected diseases. DHHS Publ No (NIH) 81-2397. Wash ington DC, US Government Printing Office, 1981. 35. Kolonel LN, Hirohata T, Nomura A. Adequacy of survey data collected from substitute respondents. Am J Epidemiol 1977; 106:476 484. 36. Rogot G, Reid DD. The validity of data from next of kin in studies of mortality among migrants. Int J Epidemiol 1975; 4:51-54.