Tobacco Documents Online

t CMiAP TER 7 The Causes of Lung Cancer In Texas PATRICIA A BUFFU:R,'' LIt*1DA WIW,WM1S PICKLE,"' THOMAS JJ MASOM" and CHAf2l..ESCONTAMT" • Epidemiology Research Unit, The University of Texas Health Science Center at Houstonj School of Public Health, Houston, Texas 77025 "Envi'ronmental Epidemiology Branch, National Cancer Institute„ I.andow,3C15, Bethesda, Maryland 20205 ABSTRACT A population-based case-comparison interview study of lung cancer was conduetedI from 1979 to 1982 in six Texas coastal counties-Orange, Jefferson, Chambers, Harris, Gal- veston, and Brazoria-to evaluate the association of lung cancer with occupational and other environmental exposures. Lung cancer mortality rates in these counties consistently haveex- eeeded lung cancer mortality rates observed for Texas and the United States from 1'950-1969 to 197U-1975 for both sexes and races (white and nonwhites). Histologically and cytologically conftmed~ incident cases diagnosed during the interval July 1976 to Jlune 1980 ~among white male and female residents aged 30-79 years were ascer- tained from,participating hospitals in the six-county area. Both population-based and de- cedent comparisons were selected and matched on age, race, sex, region of residence, and vital status at time of ascertainment. The exposures of primary interest in the study of lung cancer are those associated with oc- cupation (emplbyment in specific industries and occupations) in conjunction with tobacco,, alcohol, diet,, and residential exposures. Key Words: Smoking history„ petrochemieall industry, histologic types, construction workers, chemical! manufacturing,, transportation introdiuCfiion and Background Data presented by Doll and Peto: (1) and related reports (2) indicate that respiratory cancer sites, dominated by lung cancer, show the most dramatic in- creases of all cancer sites over the past 30 years. The roli of smoking in the etiology of respiratory cancer has been, well documented. In addition, lung cancer is @ 198A vedag Chernks Internationol;,Inc. lunp!C<ancen Causes and Preventlon:. 83

841 PbfrlCia A. BufNer, Uind© W/Mams Piixkle. Thomas J. Mason et ai recognized as possibly the most important work-related cancer. However, the in- teraction between smoking andI occupational exposures and the increased risk that may be attributed to an occupational exposure has not been very well characterized' for a large number of workplace exposures. A population-based case-comparison interview study of lung cancer, obtaining, detailed' occupational histories, was conducted in six Texas coastal counties where lung cancer mortality rates were elevated (3). Figure 1I shows the ]bcationi of'the counties of Orange, Jiefferson, Chambers, Brazoria, Galveston, and' Harris, a highly industrialized area where Houston is located. Approximately 25% (3.5 million) of the total state population in 1980 resided in, this southeastern, coastal area, the majority (77.5%)1 in Harris County. 1New11y diagnosed', histologically confirmed cases of lung cancer in white females (inrluding Hispanic),were ascertainedlfrom July 1977 through June 1'980 in Harris County (3 years) and from July 1976 through June 1!9g0 for the surrounding five counties. Similarly, cases araong white rnales (including Hispanic) were ascer tained for four years ( Jnly 1976 through June 1'9S0) for the five less urban but in- dtastrialized counties, excluding Harris County. Background lung cancer mortality rates for white males and females were examined by Texas State Economic Area Figure 1. Texaslung cancer study area.

The Causes of Lung Concen In Texas - 85 Zg re he a. .5 (al .es. ris ve. .r n- ity •ea M Top 10%JSignii > U.S. ® Low 10%aLSignif <U.S. 0 Mot Top 10%'/Signit >U.S. ® Not Low 10%/Signif < U.S:. El Not Significantly. Different From Ui.S: Figure 2. Lung cancer mortality 1!970-1975' for white malts. (SEA) for the time period (1970 to 1975') immediately preceeding, the case- comparison study. As shown in Figures 2 and 3, these maps consistently document the significantly higher hnng cancer mortality rates observed' earlier for both white males and!white females in1 these Texas: coastal counties. The dark areas along the upper Texas coast are the Beaumont SEA (Orange and Jefferson counties)„ the Houston SEA ('Harris County), and the Galveston SEA (Galveston County). Age- adjusted mortality rates (adjustedl to the 1960 Unitedl States population) in these areas are ! in the top 110 % of rates fbr SEAs in the United States and are significantly higher than the white male or white female lung cancer mortality rate for the total United States population. For white females in Harris County, this excess was notable for both the rate and the trend in the rate from 1'9'50 to 1'975 (4). For all ages, combined, the overall excess in, lung cancer mortality in the Texas study area is approximately 30-40%, but this is considerably greater for some age groups. Occupational and industrial exposures of importance for residents of the Texas coastaY area indude those associated with shipbuilding and repair, chemical and "Excluding deaths for 1972.

86 Pattkia A Buffler. Uiexio WMams Picldb, Th trn.ho s J. fvtasorn, et ol 0 ' Top 1i0%oJSignif >W!S. ®I Low 110%/Signifi< U.S. E]Nlot Top : 1'O P/o lSignif > W!S. ®I Not Low 10%/Signif <U:S. E] Mwt Sigraificantly Different From U.S' Figure 3. Lung,cancer mortality 1'970-1975' for white femalcs. petrochemical manufacturing, petroleum refining, construction, and metall in- dustries. The largest United States based chemical and synthetic rubber production facilities are located in the study area, so a high proportion of the working popula- tion currently is empioyed or has been employed'in, these industries: For some of the smaller counties, such as Otange and Jefferson, where a single industry i& dominant, as high as 27% of'the working population reported being currently employed in chemical and allied products manufacturing, compared with 2% for Harris County (5): IWlethoft Histo.ltagically confiivnedi incident't cases of lung cancer diagnosed among white male and female: residents (including Hispanic) of the study counties for the designated time intervals ( July 1977 through~ June 1980.1 for females in Harris County and July 1976 through June 1'980 for males and females in other counties) were ascertained by review of hospital l and state records. Hospitals in the study area that were not already participating in the Statewide Cancer Reporting Program V7

The Causes of'liuna Cancer irn Texas 87 were contacted and asked to participate in the study. Population-basedI and dece- dent comparison subjects were selected from state and federal'records andl matched to cases on age, race, sex, vital status at time of ascertainment, and county of residence (Harris Cbunty or other five counties). Hispanic study subjects weree identifedi systematically by use of an algorithm to identify Spanish surname. Medical records were abstractedl by state-trained abstractors to obtain relevant disease and demographic data. Follbwing, contact with the family physician (for cases only), personal interviews were conducted with study subjects or with the next of'kin of decedent cases and comparison subjects, using established criteria for selocting the most appropriate next of'kin respondents. Interviews were conducted by trained interviewers in the home using a standardized interview protoool'. Detailed information regarding the primary exposures of interest was collected~ specifically smoking history, work history, residential history, and drinking history. Industries of employment were coded'to the Standard Industrial Classification (SIC) (6) and occupations were coded using the Dictionary oODcaupational Titles (7). The Mantel-Haenszel summary chi-square and odds ratio statistics were calculated (8). Confidence intervals (95%0) were calculated using,the method of Miettinem (9): ReSW1tS' "1- in a* of is te le is s) .a. m A total of 56 of'the 67 hospitals in the six-county Texas study participated in the study, including, all of the seven large hospitals (500 or more beds). Ten of the 11 smaller hospitals that didlnot' participate were located in Harris County. Therefore we were able to ascertain 92.2% (1520 cases) of the total 1649 incident white male and female lung cancer cases (induding Hispanic) estimated fort'he 3- to 4-year in- terval (naid-1976 or 1977 to mid-1980). The number of incident cases was esti- mated mated by adjusting age-race-sex-county mortality rates by population growth and an incidence: mortality ratio of'1'.35`.1.0. Case ascertainment was higher for resi- ~ dents of counties other than~Harris County, 97.2% vs 82.1I% (Table 1). A total of j 766 female and 754 male cases were ascertained representing, respectively, 88! 7 t and 96.1 % of the total estimated incident cases ascertained. Hispanic females ap- ~ pear to be poorly ascertained :1 (38'.1%), but this may be related to the classification ~~ based on Spanish surname which may not be an effective technique for ascertain- i ing, married Hispanic females. All ascertained cases will be used for determining,age-race-sex and county lung cancer incidence rates for the study area. A total of 88.9 %p of the ascertained I cases were included in the interview study. Some cases ('11I0, or 7.2%)lacked histologic or cytologic confirmation of lung cancer and were ineligible for the case- comparison study. For the majority of these cases (79; or 71.8'%) the basis of the lung cancer diagnosis was radiologic or dinical evidence. There was insufficient diagnostic information available on the remaining 31 cases. Additional! losses of study subjects in the case-comparison study were related to race and residential eligibility criteria; unable to locate;, moved out of interview area; physician,

pattricia A. &uffNsr; Urnda WIJGIN'anns Picltle, ilhofmas J. FAason, et aV Table 1. Lung cancer case ascertainment in Texas study by sex, ethnic group, and area, 1976-1980. Number estimated Number ascertaineda (i%) Number cases interviewedb White Females Anglo 822 750 (91.2), 449 Spanish surname 42 16 (38;1) 11 Total 864 766 (88.7) 460: White Males Anglo 767 730 (95.2) 460! Spanish surname 18 24(133.3) 1!5 Total 785 754 (96.1) 475 Area Harris County 567 468 (82.1) 275 (females only, 1977-1980)1 Other counties 1082 1052 (97.2) 660 Total 1649 1520 (92.2) 935 'indudes 1110kases without histologic confirmation and an additional 1'5 cases estimated to be ineligible, in terms of rue and residence criteria. bEirdudes cases ineligible; not located;,refusab by physician, hospitali or study subject; and cases interviewed and subsequently identified as ineligible, or data to be of,pooo quality. Table 2. Texas lung,cancer study population,by sex, study groupi and ethnicity Study grouP. Cases Controls Totals Total Female 460 482 942 Male 475 466! 941! Total 935' 948' 1883 Spanish surname Female 1F 20 : 31 Male 15' 1'9 34 Total 26 39 65' hospital, and subjet:t refusals; and poor quality interview data. Overall study sub- ject refusal rates were 7.7% and 1i0.7% for decedent cases and controls respec- tively, and 1315'% and 2q! 6% for living cases and controls, respectively. A total l of 93'S interviews was completedlwith eligible cases (460 females and 475 males) and 94'8!interviews with1 freqlnenry matched comparisonsubjeets (Table 2). Irncludedlin these totals are 26 Spanish surname : cases and 39~ comparison subjects. Separate analyses are not presented at this time for these study subjects. The average duration of time study subjects resided ini the countyof diagnosis orr in the six-county study area is over 25 years for all study groups. The majority of both male ('86%) and female ('82%) 1 cases were decedent cases and were slightly older at, time of diagnosis than the living cases (Tables.3 and 4): The distribution o.f' age at diagnosis is compared for male and female study groups in Figure 4: A higher proportion of the female cases was diagnosed before. age 60 :(45'.4%) ithan male cases diagnosed before age 60 (34%). 'm ,

The Causes of Lung Cancer in Texas 89 Table 3. Number and percentage of male lung cancer cases by age:at diagnosis and type of respondent, Texas,, 1976 to 11980 err iewedb Type of respondent Total Se1f' Next of' kin Age at di i Cases Controls Cases Controls Cases Controls agnos s (years) ~ No % No % No % No % No % No % 30r39 1 1.5' 1 1.6 3 0.7 2 0.5' 4 0.8 3 0.6'. 40-49 5 7.5' 7 10.9 9 28~ 6.9 34' 8.5 33 7:0 41 8.8'. 50-59 23 34.3' 22' 34.4 102' 25.0 98 24.4 125 26.2 120: 25.7' 60}69 31 47,0' 23' 35,9 165 40:4 1164 40.8 196 41.3 183 40.2' 70-79 + 7 10.6I 11 1'7:2 1',10 27:0 1104 25.9 117 24.7 11.5' 24.7 100.0~ 10010 100:0 100.0 1I00:0 100.01 Totals 67 64 408 402' 475 466 tenns of Table 4. Number and percentage of female lung cancer cases wed I by' a8e at' diagnosis'. and l type of respondent„ Texas, 1976 to 1980. Type of respondent Total :ity Self ' Next of kin Age at di i Cases Conttols Cases Contmols Cases Controls Totals agnos s (years) No % No % No % No % No % No %. 30-39 0 0.0 3 2.6 6 11.6I 5 1.4 6' 1. 3 8 1 7 942 40-49 9 11.1 12 10.3 40 10;6I 50 1'3:7 49 10. 6' 62 . 12 9 941 50-59 36' 44.4 55 47.4 118 3I1'.1 104 28;4 1'54 33. 5 159 . 33 0 1'883 60-69 24 29;6 34' 29.3 153' 40.4 135 36;9 1'77' 38. 5 169 . 35.1 70-79 + 12 14.8 12 10.3 62' 16.4 72 1917 74 16 1 84 1'7 4'. 31 . , . 34 100.0 1:00.0 100.01 II00:0 100.0 100 0 65' Totals 81 116 379 366 460, 4821 . y sub- esp- D,ta. ,f s) and ied in _)arate )sis or mty of igbtly ion of 4'. A 1 than, Proportions of male and female cases and comparison subjects using, tobacco, cigarettes; alcohol, or who "ever lived with household member who smoked regularly" are compared in Table~ 5. Ninety-seven percent of the male cases and 91% of the female cases reported ever smoking cigarettes but a higher proportionn of the female than, male.cases reported smoking cigarettes currently, 68% vs 54%. Proportions of heavy smokers and use of alcohol (ever) were higher for cases than comparison subjects and for males than females. An extremely high proportion~ of both female cases and comparison subjects report having lived with a, household member who smoked'' regularly, 93 % vs 88117o. . Although the patterns of risk differed for males and females (Table 6), the odds ratios~ for all smoking variables were statistically significant at the p = .05 level. Among males, ex-smokers had a risk higher than, current smokers, whereas in females the risk was lo vver in ex-smokers. The highest odds ratio far females was observedl for current, smokers, 7.9 vs 5.0 for ex-smokers. Odds ratios for the ac- . tV N ?V' W ~ ~ ~ , N ~J ~ . , , ,

Patricia iA Bu(tler Undb Williams Pickle, Thomas J. Masatn et d 40.0 35:0. 30.0, 25.0• 15:0~ 1010~ 5.0 ~ 0.0. MNs Aqe atlDlaynoxis, I\`C L.yandf . 1:1 Cas.r 18.'1 17,A 40-49 5039: 6069. 70-79 yes. Femsllas Controlf FiWe 4. Age distribution (age at diagnosis) fbr male and'femalestudy subjects„Texas'lung, cancer study, 1!976-1980. Clear columns, cases; shaded! col'umns; controls: cumulated lifetime cigarette dose, expressedl as pack-years, were higher fmr males in the.low and moderate categories'but associatedl with a similar gradient in both, males and females. No difference in: risk was associatedl with the use of filtered cigarettes for either males or females. The role of "passive smoking,"' inicontributing to risk of lung,cancer was exa'rm.- inedl(Table 7). In this analysis the'crude (or unadjusted) odds ratio are increased and significant fior'both malesand'females, 1.4 and 2.1,,respectavely. However, when the confounding effect of individual subject smoking was controlled by stratifying the Table'5. Proportion of cases and controls reporting,use of tobaccoi cigarettes and alcohol by sex„ Texas lung, cancer'study, 1!97b-1980. Males Females Cases Controls Cases Controls Tobacco (K:ver)' 0.99 0.90 0.91 0:59 , Cigarettes (ever), 0.97 0.80 0.91 0.59, ' Cigarettes (current) Cigarettes (light) 0.54 0.08 0.47 0.10 0.68 0.08 0.38 0'.1'7' Cigarettes (heavy) 0.45 0.29 0.34 0.13 Alcohol (ever) 0.86 0. 811 0.78 0.61 Lived with a smoker 0.76 0.70 0.93 0.88', 41340 38:5 OR ~ v\'V 33:533.0 288 25.7 24.7'24.7 ~ .3 12.9 10:8' 1: ~. 60-69 ~. 70-79 yro.~ 30-39 30+39 40-49 50-59. _W4

The Causes of Lung Cancer irn Texas 91 Table 6. Odds ratios''associatedlwith~ smoking variables for males and! females, Texas lung,cancer study, 1976-1980 Males Females Ever smokedi 10.12 6.89: Current smoker 9.59 7.89~ Ex-smoker 10.85 5.00 Pack-years'. Low (0-35) 6.24 3.21 Moderate (36-63) 9.39' 7.98'. High ~ (6+} + ) 13'.05 13.35. Filtered cigarettes Yes 9.39 7.11 No 10.25 6.06'. Both 12.27 7.09. 'All oddd ratios:significantatp < .05. male and female study groups into smokers (ever) and nonsmokers (never) and ex- amining the adjusted odds ratios, there was no significant increase in risk associated with passive smoking. In fact, the odds ratios for nonsmokers living with a regular smoker were not increased for either males or females, 0.52 and 0. 78, resptxtively. However, odds ratios for smokers living with a regular smoker were increased, al- though not significantly, 1.28'and 1.80 for males and females. The overall odds ra- tios (adjusted) associated with passive smoking were only slightly increased and not significartt, forei'thermaleserfemales, 1.2'and11.3, respectively. When the possibility of a"`passive smoking" effect was examinedamong nonsmokers by number of years lived with a regular smoker, there was very little difference in risk for females whoo lived with a, regular smoker for 0-32 years (Table 8). The oddsratios for males sug- gest an increase by are based on smaller numbers than the analysis in, females: Table,7: Odds ratios for passive smoking,(householdlmember smoked regularly) in Texas male and'female lung, cancer studies„ 1976-1980 Yes No ~ Odds 95% Case Control Case Control ratio Confidence intervall Males Crude 361 329 93I 119 1.41' 1.04, 1.92 4L8 Self'ever smoked', No 5 56 6 34 0.52 0:15,,1.74 L2 Yes 357 273 87 85 1.28 0;91„ 1.79: 2.01 Overall (MOR) 1.20 0.87,, 1.65 11. 18'. Females Crude 429 425 24! 51 2:12' 1.29, , 3.50 9.05 Self ever smoked No 33' 164 8 32 0.78 0:34, , 1.81 0.3 Yes 396 260' 16 1!9' 1.80' 0:92„ 3.58' 3.0 Overalli(MOR) 1.30 ~ 0:78, , 2.18' 11.0 'p,< .05.

92 Patricia A Buffler, Urndm Wa'arms Pickie. Thomas J. Mosont et al Table 8. Odds ratios associated with passive smoking Ever live with household member who smoked Number Odds ratio Confidence interval. Males Total nonsmokers 6i1 0.52 0.115„ 1i.74 1.2 0-32 years 49 0.40 0.10, i'.58' 1.8 33+ years 10 1.56 0:30„ 8.05 0.3 Females Total nonsmokers 201 0.78 0.34,, 1.81 0.3 0-32 years 97 0.62 0.24, 1.63! 0.9 33+ years 99 0.93 0.38„ 2.28 0.0 Histologic types of lung cancer were classified according to the World! Health Organization (WHO) classification ('10). The four majpr cell types account for 75-85'% of the cases in both the male and female series and the cell type distribu- tion by age group is shown for males and females in Table 9. Adenocarcinom'a is the predominant lung cancer cell type in both young (30-49 years) males and females, comprising 37.8%a (males) and 38.9% (females) of all lung cancers among persons aged 30-49'years at diagnosis. There is a marked shift with age in this pat- tern such that for both males and females aged 70 or older at diagnosis the predominant cell type is squamous or epidermoid (accounting for 40.5 % of all': cases among m'ales and 31.0% among fem'ales). Overall, squamous was the predominant cell type among males (42.2%)~and adenocarcinoma among females (35.5%). These patterns held for both smokers and nonsmokers except for nonsmoking males, in whonI 6 of 11 (54.5%) cases were adenocarcinoma. The risk associated with smoking was examined by cell type, specifically odds', ratios for smoking categories within the adenocarcinorna series compared, with nonadenocarcinoma cases' (Tables 110 and 11). The odds'. ratio& for smoking categories based on pack-years were a111 significant, emphasizing the increased risk of lung cancer (all types) assfl+ciatedlwith smoking. However, the gradient of 'risk,, in both males and, females, was markedly differenr fbr adenocarcinoma compared with nonadenocarcinoma (all other lung cancer) cell types. There were 1104 cases of Table 9: Male and', female lung cancer cases by histologic type and age, Texas, 1'976-198D Males Females 30-49 years 50'-69: years 70+ years 30-49 years 50-69 years 70+ years Cell'1 type No % No % No % No % No % No % Squamous 8 2'1.6' 111'2 34.8 47 40:5 11 20.4 74! 22:6 22 31.0 Small celll 4 10.8! 64 20.1 16! 13.8 10 18.5 92 28!1 111 15.5 Adenocarcinoma 14 37.8' 73 22.9 17 14.7 21 38.9 99 30:3 19 26.8 Large cell 2 5.4 19 6.0 9' 7.8 4 7.4 11 314 3 4.2 Other 24.4 16.2 23.2 14.8 15;7 12.5' Total 100.0; 100.0 1100;0 100.0 1100,0 100.0