Montana & Mesothelioma - Materials and Methods
Participants and data collection. People were eligible for participation in the medical testing program if they had resided, worked, attended school, or participated in other activities in the Libby, Montana, area for [greater than or equal to] 6 months before 31 December 1990. We identified participants from local telephone directories and through paid newspaper, radio, and television advertisements placed locally and throughout the northwest region of the United States, and in the major newspapers in Chicago, Illinois; Boston, Massachusetts; New York, New York; and Dallas, Texas. The highest readership days of each newspaper were targeted. In addition, word of mouth and medical referrals brought participants to the testing program. A toll-free telephone line was established for interested persons to obtain information about the program and to determine eligibility for screening participants. Telephone screening to determine eligibility began in April 2000 and continued through September 2001. Eligible persons living outside of Libby had to provide their own transportation to Libby for medical evaluation. In-person interviews and medical testing were conducted in Libby in two waves: from July through November 2000 and from July through September 2001. After informed consent was obtained, trained interviewers administered a computer-assisted questionnaire to obtain demographic characteristics, residential history, occupational history, household contact history, recreational activities and other potential pathways for vermiculite exposures, cigarette smoking status, medical history, and self-reported symptoms and illnesses. The medical screening consisted of spirometric testing and chest radiographs. In this article we report on the radiographic findings of this program. Spirometric results will be reported separately.
Medical evaluation and classification of outcomes. Chest radiographs were offered to participants [greater than or equal to] 18 years of age and included posterior-anterior (P-A), right anterior-oblique, and left anterior-oblique views. Women of child-bearing age were informed that they should postpone their chest radiograph if they were pregnant. The physician on site assessed the consistency and quality of each chest radiograph taken and provided a routine radiologic interpretation, which included recording asbestos-related changes on a summary report form. If findings on a chest radiograph suggested the need for urgent medical attention, the physician completed a referral form, and the participant was counseled and directed to an appropriate source of medical care.
After initial evaluation and interpretation by the on-site physician, the participants' chest radiographs were read and classified independently by three B-readers certified by the National Institute for Occupational Safety and Health, without knowledge of the subject's age, smoking history, or exposure category, according to guidelines developed by the International Labor Organization (ILO 1980). Although the P-A view is recommended by the ILO's guidelines for the classification of radiographs (ILO 1980), we used the lateral right and left oblique views as well as the P-A view for evaluating abnormalities. The addition of right and left oblique views increases the precision and the sensitivity and specificity of chest radiographs in finding pleural abnormalities not evident on the P-A view alone (Lawson et al. 2001; Levin et al. 2000).
A subject was classified as having a pleural abnormality if two out of three B-readers indicated a) any unilateral or bilateral pleural calcification on the diaphragm, chest wall, or other site or b) any unilateral or bilateral pleural thickening or plaque on the chest wall, diaphragm, or costophrenic angle site, consistent with asbestos-related pleural disease, using the P-A view, the oblique views, or a combination of those views. A subject was classified as having an interstitial abnormality if two out of three B-readers indicated the presence of opacities, applying ILO guidelines, with a profusion of 1/0 or greater using the P-A view only.
Exposure characterization. To describe potential exposures to asbestos or vermiculite, participants were asked if they had ever worked for W.R. Grace or Zonolite (WRG) or as a contractor for WRG; if they had ever been exposed to dust at non-WRG jobs; if they had ever mixed, cut, or sprayed asbestos; if they had ever had other occupational exposure to asbestos; or if they had exposure to asbestos in military service. In addition, participants where asked if they had ever had a job as a pipe or steam fitter, plumber, brake repair person, insulator, dry wall finisher, carpenter, roofer, electrician, or welder. Some questions sought information about potential exposures in daily life, including having been a household contact of a WRG worker (during the time the worker was at WRG), length of residence in the Libby area (categorized by quartiles for analysis: 6 months through 14 years, 15-22 years, 23-34 years, and > 34 years), and having asbestos products in the home or using vermiculite in gardening or insulation (yes or no). There were questions about recreational activities such as playing in vermiculite piles, popping vermiculite, and playing on the baseball field near the expansion plant, or playing along the Rainey Creek Road--the road to the mine (never, sometimes, or frequently). Participants were also asked if they had any other contact with vermiculite (never, sometimes, or frequently).
Covariates. Age, sex, cigarette smoking status, a history of chest injury or chest surgery, tuberculosis, pneumonia, scleroderma, lupus, rheumatoid arthritis, congestive heart failure, cancer, or pulmonary disease were ascertained from the examiner-administered questionnaire. Participants were also asked if they were a) not at all, b) a little, or c) very concerned that something in the neighborhood environment may be harmful. Body mass index (BMI) was calculated from height and weight measurements obtained during the medical examination and classified into standard categories ( 18, 19-24, 25-29, and [greater than or equal to] 30) for descriptive purposes and into quartiles ( 24, 24-27, 28-31, [greater than or equal to] 32) on the basis of its distribution in the participants for statistical modeling. Age was modeled as a continuous variable, and cigarette smoking status was classified as never smoked or ever having smoked (ex-smoker and current smoker).
Statistical analysis. We used unconditional logistic regression modeling to estimate, by use of the odds ratio (OR) and 95% confidence intervals (CI), the risk of respiratory abnormalities for each of the exposure pathways while controlling for all other exposure pathways and other established and suggested risk factors for respiratory illness. Models were estimated separately for the outcomes of pleural abnormalities and interstitial abnormalities.
We first tabulated the frequencies of demographic and other participant characteristics to obtain a descriptive profile of the participants. We then tabulated prevalence of pleural abnormalities and interstitial abnormalities for each of the occupational, recreational, and household exposure pathways and for each of the covariates, as well as for the increasing number of exposure pathways. Crude ORs and 95% CIs for pleural abnormalities and interstitial abnormalities were calculated for each of these pathways and covariates. We began an unconditional logistic regression multivariate model with all potential exposure pathways and covariates. The presence of effect modification was evaluated in this model by use of interaction terms between the exposure variables and demographic characteristics. We modeled age as a continuous variable and, when appropriate, included linear and nonlinear terms in the model. Criteria for inclusion in the final model included statistical significance of the explanatory variables (p 0.10), stability of the estimate, effect on other variables in the model (confounding), biologic plausibility of the interaction, and fit of the model. We used a logistic regression model to describe an exposure-response relationship with the number of exposure pathways while controlling for covariates. A chi-square test and pairwise contrasts between pathway levels were used to assess the statistical significance of this trend. We used Procedures in SAS, release 8.01 (SAS Institute 2001) to perform all statistical analyses.
