Smoking as a confounder in case-control studies of occupational bladder cancer in women.

BACKGROUND: In studies in men, risk estimates on occupation and bladder cancer are distorted by about 10% when not adjusting for smoking. We examined the degree to which occupational risk estimates for bladder cancer in women are confounded by smoking, and the degree of residual confounding by inadequate control of this effect. METHODS: Primary data of 11 case-control studies on occupation and bladder cancer from Denmark, France, Germany, Greece, Italy, and Spain were pooled. Information for smoking and lifetime occupational history for 700 female cases and 2,425 female controls ages 30-79 was abstracted and recoded. Logistic regression was used to calculate odds ratios (OR) by occupation, applying five models which differed in their degree of adjustment for smoking. RESULTS: In major occupational groups, risk estimates were distorted by less than 10% when not adjusting for smoking. A statistically significant excess risk for bladder cancer was found in 13 specific occupations and industries. In most occupations, adjustment for smoking led the ORs towards the null value, but all statistically significant associations were maintained after adjustment. In three occupations (lathe operators, field crop workers, and wood manufacturers), a statistically significant excess risk was masked when not adjusting for smoking. In six occupations, estimates were distorted by more than 10% (-22% up to +40%). In occupations where smoking acted as a positive confounder, the proportion of confounding removed using a dichotomous smoking variable (ever/never) was around 60%. In one occupation (buyers), controlling for smoking status (ever, never) led to over-adjustment, because the percentage of smokers was high but the quantity smoked was low.     

The effects of nondifferential confounder misclassification in ecologic studies.

In ecologic studies, covariate levels of groups are often quantified as the prevalence of a dichotomous covariate. We show that, under certain conditions, nondifferential misclassification of such a binary covariate does not reduce the ability to control confounding by the covariate in ecologic studies. Thus, any remaining exposure-disease association in an adjusted ecologic analysis cannot be ascribed to incomplete control for confounding due to nondifferential misclassification of the dichotomy under those conditions, although residual confounding by the underlying covariate may still be present. This point is illustrated by ecologic analyses of the association between population density and mortality from lung cancer in women in 30 administrative districts of the Federal Republic of Germany, in which control for cigarette smoking is required.     

Smoking patterns of motor vehicle industry workers and their impact on lung cancer mortality rates

A recent retrospective mortality study, based on proportionate mortality ratios (PMRs), has shown that blue-collar automotive workers have an excess of approximately 30% in lung cancer deaths over a comparable representation of the general population. This study compares the cigarette smoking habits of automotive workers with those of the general population and assesses the extent to which any smoking habit differences might account for the excess in lung cancer deaths among automotive workers. The results show that when smoking differences are considered, the lung cancer PMR for white male automotive workers declines from 1.3 to approximately 1.1. These findings suggest that smoking habits of a studied population can affect mortality ratios for lung cancer when the smoking habits of the studied population differ significantly from those of the general population.     

Smoking prevalence and lung cancer mortality in Germany

From investigations of smoking habits, which included information about onset of smoking, we have extrapolated gender- and age-specific smoking prevalence in Germany back to the 1920s. The exploration of these data, together with an analysis of lung cancer mortality by age-period-cohort models shows that the increase of smoking and lung cancer is modulated by factors related to the country-specific socio-economic environment and that it proceeds in waves. Though recent data indicate a decline in the prevalence of smoking among both sexes, Germany is now on the way to having a similar smoking-related cancer mortality to those countries in the world which previously had a much higher rate, but have intervened vigorously. The refusal in Germany to adopt actions such as those successfully implemented to save lives in these countries means that the German population is exposed to avoidable risks, whose effects can be counted in the terms of the thousands of lives that might have been saved each year.     

Smoking as a determinant of high organochlorine levels in Greenland

The authors investigated the accumulation of organochlorines among smoking and nonsmoking Inuit hunters (n = 48) in Uummanaq, Greenland, a population with high dietary exposure to persistent organic pollutants (POPs). Human plasma organochlorine levels were positively correlated with age, marine diet, and smoking or plasma cotinine in multiple linear-regression models (p < 0.001). Body mass index was inversely correlated with organochlorine accumulation, independent of smoking status. These findings confirm that the source of POPs among the Inuit in Greenland is diet, but smoking is an important determinant of POP bioaccumulation. Smoking cessation may provide a means to lower the body burden of POPs.     

Path analysis of socioeconomic correlates of county infant mortality rates.

Relationships between selected socioeconomic characteristics of counties and infant mortality rates are examined. There are two research objectives: to determine the extent to which low family income, low education, sound housing, and the percentage of blacks "directly" and "jointly" relate to neonatal and postneonatal mortality rates; and to determine the degree to which a zero-order correlation between a given socioeconomic measure and general infant mortality is transmitted by neonatal and postneonatal mortality rates, respectively. Data corresponding to 2237 counties in the United States are analyzed by path analysis. Results show that the percentage of blacks and low education are two variables which have appreciable direct effects on both components of infant mortality. These two factors are also responsible in large measure for gross associations between low family income, sound housing, and rates of infant loss. On the basis of this study it is estimated that approximately two-thirds of the zero-order correlation between a given county measure of socioeconomic status and infant mortality occurs through the postneonatal component. Implications of these findings are discussed.     

Smoking cessation and lung cancer mortality in a cohort of middle-aged Canadian women

PURPOSE: To determine the impact of smoking cessation on lung cancer mortality among women. METHODS: Survival analysis is used to assess the effect of smoking cessation on lung cancer death in the dietary cohort of 49,165 women aged 40 to 59 years enrolled in the Canadian National Breast Screening Study. RESULTS: During an average of 10.3 years of follow-up, 106 women died of lung cancer. The risk of lung cancer mortality among women who quit before age 50 (HR=0.26; 95% CI, 0.13-0.55 among women who quit at ages 40-49) or quit in the previous 10 years (HR=0.39; 95% CI, 0.22-0.69) is substantially lower than the risk among current smokers. Women who quit after age 40 or have quit for less than 20 years are at substantially higher risk of lung cancer mortality compared with never smokers. Both duration of smoking cessation and age at quitting have independent effects on lung cancer mortality, after controlling for number of cigarettes smoked per day and number of years smoked, as well as other potential confounding variables. CONCLUSION: These findings suggest that programs and policies to promote early cessation of smoking and prevention of relapse should be a public health priority.     

Lung cancer mortality and airways obstruction among metal miners exposed to silica and low levels of radon daughters

Starting from a cross-sectional survey in 1973, the mortality of two cohorts of Sardinian metal miners was followed through December 31, 1988. In mine A, the quartz concentration in respirable dust ranged between 0.2% and 2.0% and the exposure to radon daughters averaged 0.13 working level (WL), with the highest estimated cumulative exposure around 80–120 WLM. In mine B, the silica content was much higher (6.5–29%), but exposure to radon daughters was significantly lower than in mine A. More than 98% of the overall work force in 1973 (1,741 miners) entered the cohort, providing 25,842.5 person-years. Smoking, occupational history, chest radiographs, and lung function tests were available for the cohort members at admission. Mortality for all causes was slightly lower than expected. A significant excess for nonmalignant chronic respiratory diseases was noticed in both mines. Twenty-four subjects died of lung cancer, 17 from mine A (SMR: 128; 95% confidence interval [CI]: 75–205) and 7 from mine B (SMR: 85; 95% CI: 34–175). The SMR for lung cancer was highest among the underground workers from mine A (SMR: 148; 95% CI: 74–265), with a significant upward trend by duration of employment in underground jobs. Mine B underground miners showed lung cancer SMRs close to 100 without a significant trend by duration of employment. Among underground miners with spirometric airways obstruction in 1973, those from mine A showed the highest risk (SMR: 316; 95% CI: 116–687). The relationship did not change after adjusting for age and smoking. Based on the present findings, crystalline silica per se does not appear to affect lung cancer mortality. A slight association between lung cancer mortality and exposure to radon daughters, though within relatively low levels, may be considered for underground miners from mine A. Impaired pulmonary function may be an independent predictor of lung cancer and an important risk factor enhancing the residence time of inhaled carcinogens, i.e., alpha particles or PAHs, by impairing their bronchial and alveolar clearance. © 1994 Wiley-Liss, Inc.     

Smoking in pregnancy as a risk factor for long-term mortality in the offspring

Summary. This study addresses the question of whether maternal smoking is related to postneonatal mortality and which are the contributing causes of death. Mortality of births in Sweden between 1 January 1983 and 31 December 1989 ( n = 714389) registered in the Medical Birth Registry was followed until 31 December 1990. The registry carries information on maternal smoking habits in early pregnancy. The Mantel-Haenszel procedure was used to control for confounding effects of maternal age, parity and year of birth: relative risks (95% confidence interval) associated with smoking were: 1.24 (1.17-1.31) for fetal death, 1.08 (1.01-1.16) for early neonatal death, 1.22 (1.08-1.39) for late neonatal death, 1.31 (1.23-1.41) for postneonatal mortality during the first year, and for the period between 1 and 8 years of age it was 1.19 (1.06-1.32). When birth-weight was controlled for, the increased relative risk for postneonatal mortality during the first year disappeared. Mortality between 1 and 8 years still showed an elevated risk of 1.43 (1.00–2.06). In order to control for confounding by social factors, 1986 births were linked to data from the 1985 Swedish census. The measure used was the socio-economic index (SEI), which reflects the parental education level. When maternal, paternal or family SEI was controlled for, late neonatal death was not significantly related to maternal smoking, but postneonatal death still showed a significantly increased relative risk of about 1.35. Four causes of death based on International Classification of Diseases (ICD) 8 and 9 were significantly related to maternal smoking: sudden infant death syndrome, injuries and poisoning, perinatal causes and infections.     

Empirical Bayes estimation of cancer mortality rates

Mortality rates for specific cancer types and age-sex groups computed for large numbers of cities typically show extreme fluctuation. This is primarily due to the rare occurrence of specific cancer deaths in most of the small and moderate size cities during a fixed time period. Assuming a Poisson death process, we use an empirical Bayes method to obtain adjusted rates that are more stable for comparison of cities and prediction of future mortality. We have chosen stomach and bladder cancers in Missouri cities to illustrate the problems, techniques and results.     

Brain cancer mortality rates increase with Toxoplasma gondii seroprevalence in France

The incidence of adult brain cancer was previously shown to be higher in countries where the parasite Toxoplasma gondii is common, suggesting that this brain protozoan could potentially increase the risk of tumor formation. Using countries as replicates has, however, several potential confounding factors, particularly because detection rates vary with country wealth. Using an independent dataset entirely within France, we further establish the significance of the association between T. gondii and brain cancer and find additional demographic resolution. In adult age classes 55 years and older, regional mortality rates due to brain cancer correlated positively with the local seroprevalence of T. gondii. This effect was particularly strong for men. While this novel evidence of a significant statistical association between T. gondii infection and brain cancer does not demonstrate causation, these results suggest that investigations at the scale of the individual are merited.     

Projection of incidence rates to a larger population using ecologic variables

There is wide acceptance of direct standardization of vital rates to adjust for differing age distributions according to the representation within age categories of some referent population. One can use a similar process to standardize, and subsequently project vital rates with respect to continuous, or ratio scale ecologic variables. We obtained from the National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) programme, a 10 per cent subset of the total U.S. population, country-level breast cancer incidence during 1987-1989 for white women aged 50 and over. We applied regression coefficients that relate ecologic factors to SEER incidence to the full national complement of county-level information to produce an age and ecologic factor adjusted rate that may be more representative of the U.S. than the simple age-adjusted SEER incidence. We conducted a validation study using breast cancer mortality data available for the entire U.S. and which supports the appropriateness of this method for projecting rates.     

Alcohol consumption as a major risk factor for the rise in liver cancer mortality rates in Japanese men

BACKGROUND: Age-adjusted liver cancer mortality rates have been increasing for both men and women in Japan since 1970; however, increases in mortality rates in men are much greater than those in women. Hepatitis C virus infections and heavy alcohol consumption are considered to be the major risk factors of liver cancer deaths in Japanese. The purpose of this study is (1) to examine the pattern of liver cancer mortality by gender and birth year to compare those with the pattern of other alcohol-related mortality and (2) to estimate the attributable risk per cent of heavy alcohol consumption for liver cancer deaths in Japanese men. METHODS: Age-specific liver cancer mortality rates by gender were compared with those of cirrhosis mortality rates. Then male-to-female mortality rate ratios were calculated by birth cohort and compared with cirrhosis mortality rate ratios and oesophageal cancer mortality rate ratios. The attributable risk per cent of alcohol consumption for liver cancer death was calculated, using female liver cancer mortality rates as standard rates. RESULTS: Examination of both gender and birth cohort mortality rates revealed that male-to-female liver cancer mortality rate ratios by birth cohort correspond well with those rate ratios for liver cirrhosis and oesophageal cancer mortality. The attributable risk per cent of alcohol consumption for liver cancer deaths in Japanese men was 70%. CONCLUSION: Alcohol consumption is more important than hepatitis C virus infections as a major cause of liver cancer deaths in Japanese men.     

Case-control study of lung cancer risk from residential radon exposure in Worcester county, Massachusetts

A study of lung cancer risk from residential radon exposure and its radioactive progeny was performed with 200 cases (58% male, 42% female) and 397 controls matched on age and sex, all from the same health maintenance organization. Emphasis was placed on accurate and extensive year-long dosimetry with etch-track detectors in conjunction with careful questioning about historic patterns of in-home mobility. Conditional logistic regression was used to model the outcome of cancer on radon exposure, while controlling for years of residency, smoking, education, income, and years of job exposure to known or potential carcinogens. Smoking was accounted for by nine categories: never smokers, four categories of current smokers, and four categories of former smokers. Radon exposure was divided into six categories (model 1) with break points at 25, 50, 75, 150, and 250 Bq m, the lowest being the reference. Surprisingly, the adjusted odds ratios (AORs) were, in order, 1.00, 0.53, 0.31, 0.47, 0.22, and 2.50 with the third category significantly below 1.0 (p < 0.05), and the second, fourth, and fifth categories approaching statistical significance (p < 0.1). An alternate analysis (model 2) using natural cubic splines allowed calculating AORs as a continuous function of radon exposure. That analysis produces AORs that are substantially less than 1.0 with borderline statistical significance (0.048 < or = p < or = 0.05) between approximately 85 and 123 Bq m. College-educated subjects in comparison to high-school dropouts have a significant reduction in cancer risk after controlling for smoking, years of residency, and job exposures with AOR = 0.30 (95% CI: 0.13, 0.69), p = 0.005 (model 1).     

Arsenic levels in ground water and cancer incidence in Idaho: an ecologic study

Purpose  Long-term exposure to arsenic above 50 μg/L in drinking water has been related to multiple types of cancers. Few epidemiologic studies conducted in the US have detected an association between regional exposures below this level in drinking water and corresponding cancer occurrence rates. This county-level ecologic study evaluates arsenic levels in ground water and its association with targeted cancer incidence in Idaho, where some regions have been found to contain higher arsenic levels. Methods  Using cancer incidence data (1991–2005) from the Cancer Data Registry of Idaho and arsenic data (1991–2005) from the Idaho Department of Environmental Quality, we calculated the age-adjusted incidence rate for cancers of the urinary bladder, kidney and renal pelvis, liver and bile duct, lung and bronchus, non-Hodgkin’s lymphoma (NHL), and all malignant cancers according to arsenic levels in ground water. Multivariate regression analysis was applied to evaluate the relationship between arsenic levels in ground water and cancer incidence. Results  For males, but not for females, age-adjusted incidence for lung cancer and all malignant cancers was significantly higher in the intermediate arsenic counties (2–9 μg/L, n = 16) and the high arsenic counties (≥10 μg/L, n = 5) compared to the low arsenic counties (<2.0 μg/L, n = 23). When adjusted for race, gender, population density, smoking and body mass index (BMI), no relationship was found between arsenic levels in ground water and cancer incidence. Conclusions  In this ecological design, exposure to low-level arsenic in ground water is not associated with cancer incidence when adjusting for salient variables. For populations residing in southwestern Idaho, where arsenic has been found to exceed 10 μg/L in ground water, individual risk assessment is required in order to determine whether there is a link between long-term arsenic exposure at these levels and cancer risk.     

An ecologic analysis of county-level PM2.5 concentrations and lung cancer incidence and mortality

Few studies have explored the relationship between PM2.5 and lung cancer incidence. Although results are mixed, some studies have demonstrated a positive relationship between PM2.5 and lung cancer mortality. Using an ecologic study design, we examined the county-level associations between PM2.5 concentrations (2002-2005) and lung cancer incidence and mortality in North Carolina (2002-2006). Positive trends were observed between PM2.5 concentrations and lung cancer incidence and mortality; however, the R2 for both were <0.10. The slopes for the relationship between PM2.5 and lung cancer incidence and mortality were 1.26 (95% CI 0.31, 2.21, p-value 0.01) and 0.73 (95% CI 0.09, 1.36, p-value 0.03) per 1 μg/m3 PM2.5, respectively. These associations were slightly strengthened with the inclusion of variables representing socioeconomic status and smoking. Although variability is high, thus reflecting the importance of tobacco smoking and other etiologic agents that influence lung cancer incidence and mortality besides PM2.5, a positive trend is observed between PM2.5 and lung cancer incidence and mortality. This suggests the possibility of an association between PM2.5 concentrations and lung cancer incidence and mortality.     

Epidemiological associations among lung cancer,radon exposure and elevation above sea level--a reassessment of Cohen''s county level radon study

Inhalation of radon (222Rn) decay products by persons living in homes has been associated with increased risk of lung cancer. Some epidemiological studies have shown a positive association between radon exposure and lung cancer rates. However, a large U.S.-wide ecological study (Cohen 1995) has shown a clear inverse association between average county radon concentration in homes and average lung cancer rates in the county. Cohen's strong inverse association between radon and lung cancer is surprising since there is no plausible biological reason for an inverse causal relationship between the two. We plot the county average lung cancer rate vs. the elevation above sea level (altitude) and show an inverse association between county average lung cancer rate and elevation. The elevation used for each county is the altitude of the most populous place in the county. We postulate that the decrease in lung cancer rates with higher elevations is caused by the carcinogenic effect of higher absolute oxygen concentration in the inspired air at lower elevations. Stratifying Cohen's lung cancer vs. radon data into ten groups of counties with similar elevations removes some, but not all, of his inverse association between radon and lung cancer.