Abstract

Skin lesions are classic clinical signs of toxicity due to long-term exposure to arsenic, and they are considered precursors to arsenic-related skin cancer. The authors prospectively evaluated synergisms between effects of arsenic exposure and those of tobacco use, sun exposure, and pesticide and fertilizer use on incident skin lesions using risk factor data from 5,042 men from the Health Effects of Arsenic Longitudinal Study in Araihazar, Bangladesh, which recruited participants from October 2000 to May 2002. Discrete time hazard models were used to estimate measures of synergistic interactions on the additive scale. The authors observed significant synergistic effects between various measures of arsenic exposure and smoking and fertilizer use. The relative excess risks for the interactions between smoking status and arsenic exposure were 0.12 (95% confidence interval: 0.06, 0.19) for water arsenic and 0.11 (95% confidence interval: 0.05, 0.15) for urinary arsenic measures, respectively. Significant synergistic effects were also observed between fertilizer use and water arsenic (relative excess risk for the interaction = 0.06, 95% confidence interval: 0.01, 0.12). This is the first prospective study based on individual-level data that supports a role for smoking and certain occupational risk factors in modification of the effect of long-term arsenic exposure on skin lesions. Understanding differential arsenic susceptibility allows researchers to develop interventions to prevent the health consequences of this massive problem in the Bangladeshi population and beyond.

It is estimated that approximately 100 million people worldwide have experienced long-term exposure to arsenic through naturally contaminated drinking water, with particularly high rates of exposure in Taiwan, Mexico, India, Chile, Argentina, and Bangladesh (1–10). In the 1970s, hand-pumped wells were installed in Bangladesh to provide pathogen-free groundwater for consumption. These wells gradually became the primary source of drinking water for the country's approximately 150 million people. Unfortunately, the natural arsenic contamination of the groundwater remained unknown until the 1990s and, as a result, an estimated 35–77 million people in Bangladesh have had long-term exposure to arsenic through this water supply (11–13).

Arsenic is a class 1 human carcinogen and has been linked to a variety of adverse health outcomes, including cancers of the skin, lung, bladder, liver, and kidney (3, 13–15). Arsenic-associated cancers are considered a late phenomenon related to long-term exposure to arsenic, usually occurring after ≥10 years of exposure (14). Skin lesions, however, are generally relatively early manifestations of arsenic toxicity and are often considered precursors to arsenic-induced basal and squamous cell skin cancers (16). Therefore, understanding the risk factors that contribute to skin lesion incidence due to arsenic may improve our understanding of the cause of arsenic-related skin cancers.

Prior studies have suggested that arsenic exposure is necessary for the development of premalignant skin lesions; however, high interindividual variability and differential arsenic susceptibility in men and women suggest that perhaps it alone is not sufficient to cause lesions (17). Numerous epidemiologic studies have examined the relation between long-term arsenic exposure and the prevalence of skin lesions in populations worldwide. Cross-sectional and retrospective studies conducted in Bangladesh (7, 18) (M. Argos, University of Chicago, unpublished data, 2010), India (4), and Chile (6) have suggested significant trends for the prevalence rates of skin lesions in relation to arsenic exposure status, as well as differing rates of susceptibility in men and women.

The Health Effects of Arsenic Longitudinal Study (HEALS) was initiated in 2000 to prospectively investigate the relation between long-term arsenic exposure and adverse health outcomes, with an initial focus on the effects of arsenic exposure at low to moderate levels on the incidence of premalignant skin lesions. We previously evaluated the modifying effects of sun exposure, tobacco use, and pesticide and fertilizer use on the relation between arsenic and skin lesion prevalence in a cross-sectional analysis of baseline HEALS data (16). Cross-sectional findings suggested an apparent synergistic effect between a high level of arsenic exposure and tobacco smoking in men (16, 19). Whereas the effects of smoking on internal organs have been studied extensively and associations between smoking and cancers of the lung, bladder, esophagus, and cervix have been well-established, studies have not consistently identified smoking as a significant risk factor for nonmelanoma skin cancers (20). In the present study, we utilized 6 years’ worth of follow-up data on the HEALS cohort to prospectively evaluate synergistic interactions between long-term arsenic exposure and environmental and occupational factors, especially smoking, in relation to skin lesion incidence.

MATERIALS AND METHODS

The Health Effects of Arsenic Longitudinal Study

HEALS is a prospective cohort study of a population-based sample of adults in Araihazar, Bangladesh. Detailed information on study methodology and sampling has been previously described (18). In short, a sampling frame was created by collecting water samples, geographic data, and basic demographic information for the users of 5,966 contiguous wells in a 25-km2 study area (5). Between October 2000 and May 2002, a total of 11,746 married individuals aged 18–75 years were recruited for inclusion in the cohort, 5,042 of whom were men. Because exposure to sun, tobacco, pesticides, and fertilizers is uncommon among women in this study population, we restricted our present study to men only.

Trained research physicians conducted baseline interviews that included questions about water-drinking patterns and history, as well as about demographic information and lifestyle characteristics pertaining to occupation and diet. In addition to the interview data, study personnel collected urine and blood samples from participants in their homes according to a structured protocol. Trained physicians conducted a comprehensive physical examination, with a specific focus on the signs and symptoms of arsenic-related diseases, including skin lesions and skin cancers. Participants were subsequently visited for biennial follow-up evaluations, which consisted of an interviewer-administered questionnaire, a physician-administered clinical examination, and urine sample collection. Study physicians were blinded to the arsenic level of the participants’ drinking water during all visits.

Exposure assessment

Three primary measures of chronic arsenic exposure were obtained at baseline for each individual in the HEALS cohort: well-water arsenic concentration, daily arsenic intake from drinking water, and creatinine-adjusted urinary total arsenic concentration. At baseline, participants were asked to identify the well used as their primary source of drinking water. Corresponding well-water arsenic exposure levels were subsequently assigned based on these responses. The concentrations of arsenic in each of the 5,966 tube wells assessed in the study were measured using laboratory-based methods previously described (16, 18). In short, water arsenic concentrations were measured using graphite furnace atomic absorption spectrometry with a detection limit of 5 μg/L. Samples below the limit of detection were subsequently reanalyzed using inductively coupled plasma mass spectrometry with a detection limit of 0.1 μg/L (16, 18). Daily arsenic intake (μg/day) was calculated by multiplying the well-water arsenic concentration of the primary well (μg/L) by the self-reported daily amount consumed from that well (L/day).

Out of the total male cohort (n = 5,042), approximately 96% (n = 4,855) provided a spot urine sample at baseline. Urinary total arsenic concentration was measured using graphite furnace atomic absorption spectrometry with a detection limit of 2 μg/L. Urinary creatinine was measured using a colorimetric Sigma Diagnostics Kit (Sigma-Aldrich Corporation, St. Louis, Missouri), and the urinary total arsenic value was subsequently divided by creatinine level to obtain the value for creatinine-adjusted urinary total arsenic concentration (expressed as μg/g creatinine) (18, 21).

Well-water arsenic cutpoints for the first and second well-water quintiles at baseline corresponded to the World Health Organization's guidelines for arsenic in drinking water (≤10 μg/L) and the national standard for arsenic in drinking water in Bangladesh (≤50 μg/L), respectively. Urinary total arsenic concentration and daily arsenic dose were divided into quintiles on the basis of the baseline distribution of the cohort.

Of the 5,042 male participants with baseline sample data, we excluded 584 individuals who had prevalent skin lesions at baseline, 165 who had missing data on baseline skin examination, and 302 who had missing data on follow-up skin examination. Thus, 3,991 men were eligible and included in the current analysis using well-water arsenic concentrations. Of those, 3,804 men had provided spot urine samples; corresponding analysis for urinary arsenic concentration was conducted in that group.

Outcome assessment

Methods for the ascertainment of skin lesions in HEALS have been described in detail elsewhere (11, 18). Briefly, using a structured protocol, trained study physicians examined each subject and recorded the presence or absence of skin lesions, the location of any skin lesions, and their size and shape at baseline and each biennial follow-up visit. During the study follow-up period, 613 incident skin lesions were detected in the study sample. Participants who did not develop skin lesions were censored at the third biennial follow-up (n = 3,173) or at the time of last skin examination (n = 205). For the purposes of these analyses, once an individual was censored, his data was not reentered into the analysis cohort.

Occupational and lifestyle factors

To evaluate possible interactions between arsenic exposure and other environmental or occupational covariates, we collected detailed information about potential risk factors. These risk factors included smoking status and occupational uses of fertilizers and pesticides, including additional information regarding duration and current use. Men with outdoor occupations were queried about daytime working hours and whether their bodies remained covered while working outside to obtain an approximate measurement of the extent of sun exposure.

For the purpose of this analysis, participants were classified as past, current, or never smokers. We further categorized smoking status by calculating pack-years as the product of cigarettes per day and years of smoking divided by 20. Pack-year categories of no exposure, low exposure, and high exposure were created based on the median of the distribution of pack-years among those with any exposure in the total male cohort population (10 pack-years).

Similar to the categorization of pack-years, low and high fertilizer use, pesticide use, and sun exposure categories were constructed based on the median value of the number of hours exposed per day in the male cohort population among those with exposure. The median exposure times for those participants who answered yes to fertilizer use, pesticide use, and sun exposure were 10 years, 7 years, and 8 hours per day, respectively.

Covariates

Covariate data were derived from the baseline interview. Sociodemographic factors included age (continuous years), years of education (continuous), and whether the participant received formal education (yes or no). Body mass index (calculated as weight in kilograms divided by height in meters squared) was derived from measured height and weight at baseline. Standard international cutoffs were used to classify participants into underweight (<18.5), healthy weight (18.5–24.9), and overweight (≥25) categories.

Statistical analysis

Discrete time hazard models were used to estimate discrete time hazard ratios and their 95% confidence intervals for incidence of skin lesions. Discrete time hazard was defined in this study as the probability that an individual would develop a skin lesion in each biennial follow-up cycle, conditional on that individual being lesion-free in the previous study interval (15). Conditional probability of a new lesion was estimated using a log-linear model with common regression coefficients across all intervals, with different intercepts for each. Regression coefficients are log discrete time hazard ratios, analogous to continuous time hazard ratios corresponding to traditional proportional hazards models (22). We used robust standard errors for discrete time hazards to account for potential correlation due to the clustered sampling of individuals within each primary well group (23, 24).

Arsenic exposure quintiles (for urinary total arsenic, daily arsenic intake, and water arsenic concentrations) were first modeled as dummy variables in separate regression models adjusting for age. Body mass index, formal education, years of education, and follow-up interval were included together in the analysis as potential confounders based on our a priori understanding of causal factors associated with arsenic-related skin lesions. Models were further adjusted for environmental covariates based on their interrelationships in our study population. Specifically, models for the effects of fertilizer use, pesticide use, and sun exposure were adjusted for the remaining covariates because of high intervariable correlation. Models assessing smoking effects (both smoking status and pack-years) were not adjusted for these additional environmental or occupational modifiers, as those variables were not found to be correlated with smoking variables in our study population.

In certain biologic models, the joint effects of 2 factors follow an additive pattern on the hazard of the outcome of interest (25). Therefore, in this analysis, we assessed the presence of interaction on the additive scale between arsenic exposure and each of the potential interacting factors—sun exposure, fertilizer use, pesticide use, and smoking status. Multivariate-adjusted estimates were used to estimate the relative excess risk for interaction (RERI), which was calculated as follows: 

graphic
where β1 is the coefficient of the ordinal arsenic exposure measure, β2 is the coefficient of the ordinal effect modifier measure, and β3 is the coefficient of the cross-product of the ordinal arsenic exposure and ordinal effect modifier (25, 26). Bias-corrected and accelerated 95% confidence intervals of the RERI were estimated via 1,000 bootstrap samples, in which the resampling was performed at the level of the well. Estimates for 95% confidence intervals were also calculated using the standard delta method described by Hosmer and Lemeshow, with similar results (not shown) (27). Statistical analyses were performed using the Statistical Analysis System, including the procedure GENMOD for analysis, release 9.2 (SAS Institute, Inc., Cary, North Carolina), and STATA, version 11 (StataCorp LP, College Station, Texas).

RESULTS

In the 3,991 men in this analysis, a total of 613 incident skin lesions were detected during the 6-year follow-up period. Characteristics of the cohort according to skin lesion status are shown in Table 1. Dose-dependent associations for skin lesion incidence were seen with well-water arsenic and urinary total arsenic concentrations. In unadjusted models, having a formal education, increasing years of formal education, and body mass index were associated with decreased risk of skin lesions, whereas the remaining covariates (fertilizer use, pesticide use, increasing age, current smoking, and sun exposure) were associated with increased skin lesion incidence. The distribution of daily arsenic intake was qualitatively similar to the distribution of well-water arsenic concentration and is therefore not shown in Table 1.

Table 1.

Characteristics of Participants in the Health Effects of Arsenic Longitudinal Study in Relation to Skin Lesion Status, 2000–2006

Characteristic Developed Skin Lesions (n = 613)
 
No Skin Lesions (n = 3,378)
 
HRa 95% CI P for Trend 
No. No. 
Well-water arsenic, μg/L       <0.001 
    0.1–10 112 18 910 27 1.00   
    10.1–50 90 15 784 23 1.08 0.85, 1.38  
    50.1–100 114 19 594 18 1.52 1.19, 1.93  
    100.1–200 228 37 639 19 1.86 1.48, 2.32  
    ≥200.1 69 11 449 13 2.69 2.16, 3.35  
Urinary total arsenic, μg/g       <0.001 
    ≤89 112 19 854 26 1.00   
    89.1–159 94 16 748 23 0.89 0.70, 1.14  
    159.1–245 128 21 712 22 1.28 1.02, 1.60  
    245.1–405 133 22 584 18 1.38 1.10, 1.73  
    >405 137 23 414 13 1.86 1.50, 2.31  
Body mass index, kg/m2       <0.001 
    <18.5 285 47 1,451 43 1.00   
    18.5–24.9 304 50 1,684 50 0.83 0.72, 0.95  
    ≥25 21 231 0.47 0.34, 0.67  
Education, years       0.002 
    0 285 47 1,271 38 1.00   
    1–5 304 50 1,017 30 0.85 0.73, 0.99  
    ≥6 21 1,085 32 0.76 0.64, 0.91  
Age, years       <0.001 
    18–30 29 587 17 1.00   
    31–40 138 23 1,427 42 2.91 2.23, 3.78  
    41–50 229 37 883 26 6.54 5.09, 8.50  
    51–75 217 35 481 14 13.50 10.29, 17.71  
Fertilizer use, years       <0.001 
    0 4,163 43 1,170 35 1.00   
    <10 (low) 4,861 50 1,061 31 1.97 1.65, 2.35  
    ≥10 (high) 648 1,147 34 4.03 3.46, 4.69  
Pesticide use, years       <0.001 
    0 222 36 1,702 51 1.00   
    <7 (low) 138 23 801 24 2.43 2.03, 2.91  
    ≥7 (high) 250 41 864 26 4.05 3.47, 4.71  
Working outside uncovered       <0.001 
    No outside work 390 64 2,597 77 1.00   
    Works outside covered 174 28 644 19 3.24 2.75, 3.83  
    Works outside uncovered 49 137 4.13 3.10, 5.49  
Time worked outside, hours/day       <0.001 
    0 389 63 2,593 77 1.00   
    <8 (low) 106 17 358 11 3.43 2.80, 4.19  
    ≥8 (high) 118 19 427 13 3.33 2.73, 4.06  
Cigarette smoking       <0.001 
    Never 109 18 983 29 1.00   
    Former 122 20 332 10 5.12 4.20, 6.22  
    Current 382 62 2,062 61 3.39 2.93, 3.92  
Pack-years       <0.001 
    Nonsmoker 109 18 984 29 1.00   
    <10 years (low) 170 28 1,062 31 2.74 2.30, 3.27  
    ≥10 years (high) 328 54 1,321 39 4.61 3.97, 5.36  
Characteristic Developed Skin Lesions (n = 613)
 
No Skin Lesions (n = 3,378)
 
HRa 95% CI P for Trend 
No. No. 
Well-water arsenic, μg/L       <0.001 
    0.1–10 112 18 910 27 1.00   
    10.1–50 90 15 784 23 1.08 0.85, 1.38  
    50.1–100 114 19 594 18 1.52 1.19, 1.93  
    100.1–200 228 37 639 19 1.86 1.48, 2.32  
    ≥200.1 69 11 449 13 2.69 2.16, 3.35  
Urinary total arsenic, μg/g       <0.001 
    ≤89 112 19 854 26 1.00   
    89.1–159 94 16 748 23 0.89 0.70, 1.14  
    159.1–245 128 21 712 22 1.28 1.02, 1.60  
    245.1–405 133 22 584 18 1.38 1.10, 1.73  
    >405 137 23 414 13 1.86 1.50, 2.31  
Body mass index, kg/m2       <0.001 
    <18.5 285 47 1,451 43 1.00   
    18.5–24.9 304 50 1,684 50 0.83 0.72, 0.95  
    ≥25 21 231 0.47 0.34, 0.67  
Education, years       0.002 
    0 285 47 1,271 38 1.00   
    1–5 304 50 1,017 30 0.85 0.73, 0.99  
    ≥6 21 1,085 32 0.76 0.64, 0.91  
Age, years       <0.001 
    18–30 29 587 17 1.00   
    31–40 138 23 1,427 42 2.91 2.23, 3.78  
    41–50 229 37 883 26 6.54 5.09, 8.50  
    51–75 217 35 481 14 13.50 10.29, 17.71  
Fertilizer use, years       <0.001 
    0 4,163 43 1,170 35 1.00   
    <10 (low) 4,861 50 1,061 31 1.97 1.65, 2.35  
    ≥10 (high) 648 1,147 34 4.03 3.46, 4.69  
Pesticide use, years       <0.001 
    0 222 36 1,702 51 1.00   
    <7 (low) 138 23 801 24 2.43 2.03, 2.91  
    ≥7 (high) 250 41 864 26 4.05 3.47, 4.71  
Working outside uncovered       <0.001 
    No outside work 390 64 2,597 77 1.00   
    Works outside covered 174 28 644 19 3.24 2.75, 3.83  
    Works outside uncovered 49 137 4.13 3.10, 5.49  
Time worked outside, hours/day       <0.001 
    0 389 63 2,593 77 1.00   
    <8 (low) 106 17 358 11 3.43 2.80, 4.19  
    ≥8 (high) 118 19 427 13 3.33 2.73, 4.06  
Cigarette smoking       <0.001 
    Never 109 18 983 29 1.00   
    Former 122 20 332 10 5.12 4.20, 6.22  
    Current 382 62 2,062 61 3.39 2.93, 3.92  
Pack-years       <0.001 
    Nonsmoker 109 18 984 29 1.00   
    <10 years (low) 170 28 1,062 31 2.74 2.30, 3.27  
    ≥10 years (high) 328 54 1,321 39 4.61 3.97, 5.36  

Abbreviations: CI, confidence interval; HR, hazard ratio.

a

HR represents a discrete time hazard ratio.

The results from the models assessing interactions between well-water arsenic concentration and occupational and environmental effect modifiers are reported in Table 2. In multivariate adjusted models, the association between arsenic and skin lesions was generally stronger in the presence of each candidate effect modifier. For example, in all 3 categories of fertilizer use, we saw dose-dependent increases in the risk of skin lesions by each quintile of water arsenic exposure. Similarly, at each quintile of water arsenic exposure, the risk of skin lesions was consistently higher for those with either level of fertilizer use (above or below mean exposure time of 10 years). This pattern was present for each of the effect modifiers in question, with particularly large effect sizes evident across categories of smoking status, with a hazard ratio of 10.25 (95% confidence interval: 6.48, 16.23) for past smokers with the highest level of arsenic exposure and a hazard ratio of 8.35 (95% confidence interval: 5.53, 12.61) for current smokers with the highest level of arsenic exposure when compared with nonsmoking individuals at the lowest level of arsenic exposure.

Table 2.

Risk of Skin Lesions in Relation to Levels of Well-Water Arsenic Concentration, by Fertilizer Use, Pesticide Use, Sun Exposure, and Smoking Status, in the Health Effects of Arsenic Longitudinal Study, 2000–2006

Exposure Status Well-Water Arsenic Level, μg/L
 
0.1–10
 
10.1–50
 
50.1–100
 
100.1–200
 
≥200.1
 
RERI 95% CI 
HR 95% CI HR 95% CI HR 95% CI HR 95% CI HR 95% CI 
Fertilizer use,a years             
    0 1.00  0.70 0.37, 1.30 1.11 0.64, 1.93 1.35 0.82, 2.22 2.18 1.35, 3.53 0.06 0.01, 0.12 
    <10 (low) 1.06 0.66, 1.74 0.81 0.47, 1.40 1.45 0.85, 2.47 2.11 1.34, 3.33 2.06 1.27, 3.34   
    ≥10 (high) 0.94 0.60, 1.49 1.6 0.80, 2.00 1.82 1.15, 2.88 1.82 1.15, 2.88 3.06 1.95, 4.79   
Pesticide use,b years             
    0 1.00  0.62 0.39, 1.02 1.21 0.79, 1.87 1.35 0.91, 1.98 2.33 1.63, 3.32 0.01 −0.04, 0.06 
    <7 (low) 0.67 0.41, 1.11 0.84 0.51, 1.38 1.55 0.99, 2.43 1.99 1.33, 2.97 2.07 1.34, 3.19   
    ≥7 (high) 0.85 0.54, 1.34 1.11 0.74, 1.69 1.38 0.90, 2.10 1.50 0.98, 2.30 2.22 1.49, 3.31   
Sun exposurec             
    No outdoor work 1.00  0.77 0.53, 1.10 1.52 1.10, 2.09 1.68 1.22, 2.28 2.66 1.99, 3.56 0.01 −0.06, 0.09 
    Works outdoors, covered 0.96 0.63, 1.48 1.68 1.15, 2.45 1.75 1.16, 2.64 1.85 1.28, 2.67 2.30 1.54, 3.41   
    Works outdoors, uncovered 1.17 0.59, 2.30 1.38 0.66, 2.86 0.99 0.48, 2.08 3.16 1.82, 5.50 2.97 1.71, 5.16   
Time worked outside,c hours/day             
    0 1.00  0.77 0.53, 1.10 1.52 1.11, 2.10 1.68 1.23, 2.27 2.63 1.97, 3.53 0.02 −0.04, 0.08 
    <8 (low) 1.14 0.71, 1.82 1.49 0.99, 2.27 1.44 0.86, 2.39 1.60 1.03, 2.47 2.22 1.35, 3.67   
    ≥8 (high) 1.14 0.11, 1.82 1.50 0.92, 2.46 1.82 1.15, 2.89 2.62 1.73, 3.97 2.67 1.74, 4.08   
Smokingd             
    Never 1.00  1.72 1.09, 2.71 2.55 1.64, 3.97 3.07 2.03, 4.66 4.84 3.21, 7.30 0.12 0.06, 0.19 
    Past 4.44 2.62, 7.51 3.83 2.20, 6.70 4.94 2.90, 8.42 6.67 4.06, 10.95 10.25 6.48, 16.23   
    Current 3.66 2.40, 5.59 3.80 2.48, 5.82 5.40 3.51, 8.29 5.87 3.88, 8.88 8.35 5.53, 12.61   
Pack-yearsd             
    0 1.00  1.72 1.10, 2.71 3.07 2.02, 4.65 2.55 1.64, 3.96 4.84 3.27, 7.29 0.15 0.09, 0.24 
    <10 (low) 3.01 1.86, 4.87 3.11 1.91, 5.09 3.92 2.37, 6.49 5.90 3.75, 9.28 8.11 5.18, 12.69   
    ≥10 (high) 4.38 2.85, 6.73 4.44 2.85, 6.90 6.30 4.08, 9.73 8.95 5.86, 13.67 6.34 4.15, 9.68   
Exposure Status Well-Water Arsenic Level, μg/L
 
0.1–10
 
10.1–50
 
50.1–100
 
100.1–200
 
≥200.1
 
RERI 95% CI 
HR 95% CI HR 95% CI HR 95% CI HR 95% CI HR 95% CI 
Fertilizer use,a years             
    0 1.00  0.70 0.37, 1.30 1.11 0.64, 1.93 1.35 0.82, 2.22 2.18 1.35, 3.53 0.06 0.01, 0.12 
    <10 (low) 1.06 0.66, 1.74 0.81 0.47, 1.40 1.45 0.85, 2.47 2.11 1.34, 3.33 2.06 1.27, 3.34   
    ≥10 (high) 0.94 0.60, 1.49 1.6 0.80, 2.00 1.82 1.15, 2.88 1.82 1.15, 2.88 3.06 1.95, 4.79   
Pesticide use,b years             
    0 1.00  0.62 0.39, 1.02 1.21 0.79, 1.87 1.35 0.91, 1.98 2.33 1.63, 3.32 0.01 −0.04, 0.06 
    <7 (low) 0.67 0.41, 1.11 0.84 0.51, 1.38 1.55 0.99, 2.43 1.99 1.33, 2.97 2.07 1.34, 3.19   
    ≥7 (high) 0.85 0.54, 1.34 1.11 0.74, 1.69 1.38 0.90, 2.10 1.50 0.98, 2.30 2.22 1.49, 3.31   
Sun exposurec             
    No outdoor work 1.00  0.77 0.53, 1.10 1.52 1.10, 2.09 1.68 1.22, 2.28 2.66 1.99, 3.56 0.01 −0.06, 0.09 
    Works outdoors, covered 0.96 0.63, 1.48 1.68 1.15, 2.45 1.75 1.16, 2.64 1.85 1.28, 2.67 2.30 1.54, 3.41   
    Works outdoors, uncovered 1.17 0.59, 2.30 1.38 0.66, 2.86 0.99 0.48, 2.08 3.16 1.82, 5.50 2.97 1.71, 5.16   
Time worked outside,c hours/day             
    0 1.00  0.77 0.53, 1.10 1.52 1.11, 2.10 1.68 1.23, 2.27 2.63 1.97, 3.53 0.02 −0.04, 0.08 
    <8 (low) 1.14 0.71, 1.82 1.49 0.99, 2.27 1.44 0.86, 2.39 1.60 1.03, 2.47 2.22 1.35, 3.67   
    ≥8 (high) 1.14 0.11, 1.82 1.50 0.92, 2.46 1.82 1.15, 2.89 2.62 1.73, 3.97 2.67 1.74, 4.08   
Smokingd             
    Never 1.00  1.72 1.09, 2.71 2.55 1.64, 3.97 3.07 2.03, 4.66 4.84 3.21, 7.30 0.12 0.06, 0.19 
    Past 4.44 2.62, 7.51 3.83 2.20, 6.70 4.94 2.90, 8.42 6.67 4.06, 10.95 10.25 6.48, 16.23   
    Current 3.66 2.40, 5.59 3.80 2.48, 5.82 5.40 3.51, 8.29 5.87 3.88, 8.88 8.35 5.53, 12.61   
Pack-yearsd             
    0 1.00  1.72 1.10, 2.71 3.07 2.02, 4.65 2.55 1.64, 3.96 4.84 3.27, 7.29 0.15 0.09, 0.24 
    <10 (low) 3.01 1.86, 4.87 3.11 1.91, 5.09 3.92 2.37, 6.49 5.90 3.75, 9.28 8.11 5.18, 12.69   
    ≥10 (high) 4.38 2.85, 6.73 4.44 2.85, 6.90 6.30 4.08, 9.73 8.95 5.86, 13.67 6.34 4.15, 9.68   

Abbreviations: CI, confidence interval; HR, hazard ratio; RERI, relative excess risk due to interaction.

a

Adjusted for pesticide use, working uncovered, smoking, body mass index, formal education, years of education, and follow-up interval.

b

Adjusted for fertilizer use, working uncovered, smoking, body mass index, formal education, years of education, and follow-up interval.

c

Adjusted for fertilizer use, pesticide use, smoking, body mass index, formal education, years of education, and follow-up interval.

d

Adjusted for body mass index, formal education, years of education, and follow-up interval.

We evaluated whether there was interaction (on the additive scale) between water arsenic concentration and each of the occupational and environmental effect modifiers (fertilizer use, pesticide use, smoking status, and sun exposure) in relation to skin lesion incidence. We observed that the joint effects of well-water arsenic concentration and exposure to certain occupational and environmental risk factors were greater than the sum of their individual effects. For example, the calculated RERI for the interaction between smoking status and water arsenic exposure was 0.12 (95% confidence interval: 0.06, 0.19), indicating that for every quintile increase in arsenic exposure and unit increase in fertilizer exposure, the hazard ratio was 0.12 more than if there was no interaction (i.e., if the risks were additive). Analysis of smoking status, pack years, and fertilizer use yielded evidence of a synergistic (additive) interaction for water arsenic exposure.

We also analyzed the potential interactions between environmental and occupational risk factors and creatinine-adjusted urinary total arsenic concentrations (Table 3). In general, the estimates for risk of skin lesions in relation to increasing levels of arsenic exposure or increasing levels of environmental or occupational exposure were similar to those seen in Table 2. However, RERI estimates were slightly attenuated, and analyses of urinary total arsenic concentration on risk of skin lesions yielded nonsignificant results for fertilizer use and statistically significant synergistic effects for smoking status and pack-years only.

Table 3.

Risk of Skin Lesions in Relation to Levels of Urinary Arsenic Concentration, by Fertilizer Use, Pesticide Use, Sun Exposure and Smoking Status, in the Health Effects of Arsenic Longitudinal Study, 2000–2006

Exposure Status Urinary Arsenic Concentration, μg/g
 
≤89
 
89.1–159
 
159.1–245
 
245.1–405
 
>405
 
RERI 95% CI 
HR 95% CI HR 95% CI HR 95% CI HR 95% CI HR 95% CI 
Fertilizer use,a years             
    0 1.00  0.72 0.41, 1.25 0.98 0.71, 1.68 1.40 0.86, 2.29 1.85 1.11, 3.09 0.04 −0.01, 0.10 
    <10 (low) 1.17 0.64, 2.12 0.84 0.50, 1.43 1.30 0.81, 2.08 1.81 1.12, 2.93 2.07 1.28, 3.32   
    ≥10 (high) 1.03 0.66, 1.61 1.15 0.72, 1.83 1.38 0.82, 2.18 1.61 1.02, 2.54 2.60 1.65, 4.08   
Pesticide use,b years             
    0 1.00  0.73 0.47, 1.13 1.07 0.71, 1.62 1.55 1.05, 2.30 2.16 1.47, 3.18 −0.01 −0.06, 0.05 
    <7 (low) 0.73 0.44, 1.23 2.63 1.71, 4.05 1.50 0.99, 2.29 1.46 0.92, 2.30 0.93 0.57, 1.52   
    ≥7 (high) 0.89 0.52, 1.53 1.16 0.70, 1.91 1.44 0.87, 2.41 1.58 0.94, 2.65 2.33 1.43, 3.80   
Sun exposurec             
    No outdoor work 1.00  0.90 0.65, 1.25 1.48 1.08, 2.02 1.25 0.92, 1.69 2.50 1.85, 3.37   
    Works outdoors, covered 1.07 0.68, 1.71 1.18 0.76, 1.81 1.35 0.91, 1.99 2.02 1.41, 2.88 2.31 1.58, 3.38   
    Works outdoors, uncovered 1.39 0.72, 2.68 1.22 0.61, 2.42 1.83 1.03, 3.26 1.99 1.08, 3.65 1.78 0.91, 3.50   
Time worked outside,c hours/day             
    0 1.00  0.89 0.65, 1.25 1.24 0.92, 1.69 2.50 1.85, 3.37 1.47 1.07, 2.01 0.02 −0.04, 0.08 
    <8 (low) 1.40 0.89, 2.23 0.67 0.36, 1.26 0.93 0.54, 1.63 1.49 0.89, 2.49 1.17 0.64, 2.15   
    ≥8 (high) 0.88 0.47, 1.65 1.63 0.78, 3.40 1.77 0.88, 3.71 2.19 1.10, 4.39 3.15 1.59, 6.25   
Smokingd             
    Never 1.00  1.13 0.73, 1.75 1.71 1.14, 2.57 1.80 1.20, 2.68 2.94 2.01, 4.29 0.11 0.05, 0.15 
    Past 3.96 2.44, 6.44 3.22 1.94, 5.37 4.12 2.55, 6.67 3.87 2.33, 6.44 5.53 3.36, 9.10   
    Current 2.67 1.78, 4.02 2.32 1.53, 3.54 3.18 2.15, 4.74 4.46 3.03, 6.58 5.87 3.98, 8.68   
Pack-yearsd             
    0 1.00  1.13 0.73, 1.75 1.71 1.14, 2.57 1.79 1.20, 2.68 2.93 2.00, 4.29 0.12 0.06, 0.19 
    <10 (low) 2.40 1.51, 3.81 1.73 1.04, 2.89 3.30 2.11, 5.17 3.96 2.56, 6.13 4.95 3.20, 7.65   
    ≥10 (high) 3.34 2.21 5.06 3.17 2.07, 4.85 3.48 2.35, 5.23 4.54 3.03, 6.82 6.48 4.33, 9.72   
Exposure Status Urinary Arsenic Concentration, μg/g
 
≤89
 
89.1–159
 
159.1–245
 
245.1–405
 
>405
 
RERI 95% CI 
HR 95% CI HR 95% CI HR 95% CI HR 95% CI HR 95% CI 
Fertilizer use,a years             
    0 1.00  0.72 0.41, 1.25 0.98 0.71, 1.68 1.40 0.86, 2.29 1.85 1.11, 3.09 0.04 −0.01, 0.10 
    <10 (low) 1.17 0.64, 2.12 0.84 0.50, 1.43 1.30 0.81, 2.08 1.81 1.12, 2.93 2.07 1.28, 3.32   
    ≥10 (high) 1.03 0.66, 1.61 1.15 0.72, 1.83 1.38 0.82, 2.18 1.61 1.02, 2.54 2.60 1.65, 4.08   
Pesticide use,b years             
    0 1.00  0.73 0.47, 1.13 1.07 0.71, 1.62 1.55 1.05, 2.30 2.16 1.47, 3.18 −0.01 −0.06, 0.05 
    <7 (low) 0.73 0.44, 1.23 2.63 1.71, 4.05 1.50 0.99, 2.29 1.46 0.92, 2.30 0.93 0.57, 1.52   
    ≥7 (high) 0.89 0.52, 1.53 1.16 0.70, 1.91 1.44 0.87, 2.41 1.58 0.94, 2.65 2.33 1.43, 3.80   
Sun exposurec             
    No outdoor work 1.00  0.90 0.65, 1.25 1.48 1.08, 2.02 1.25 0.92, 1.69 2.50 1.85, 3.37   
    Works outdoors, covered 1.07 0.68, 1.71 1.18 0.76, 1.81 1.35 0.91, 1.99 2.02 1.41, 2.88 2.31 1.58, 3.38   
    Works outdoors, uncovered 1.39 0.72, 2.68 1.22 0.61, 2.42 1.83 1.03, 3.26 1.99 1.08, 3.65 1.78 0.91, 3.50   
Time worked outside,c hours/day             
    0 1.00  0.89 0.65, 1.25 1.24 0.92, 1.69 2.50 1.85, 3.37 1.47 1.07, 2.01 0.02 −0.04, 0.08 
    <8 (low) 1.40 0.89, 2.23 0.67 0.36, 1.26 0.93 0.54, 1.63 1.49 0.89, 2.49 1.17 0.64, 2.15   
    ≥8 (high) 0.88 0.47, 1.65 1.63 0.78, 3.40 1.77 0.88, 3.71 2.19 1.10, 4.39 3.15 1.59, 6.25   
Smokingd             
    Never 1.00  1.13 0.73, 1.75 1.71 1.14, 2.57 1.80 1.20, 2.68 2.94 2.01, 4.29 0.11 0.05, 0.15 
    Past 3.96 2.44, 6.44 3.22 1.94, 5.37 4.12 2.55, 6.67 3.87 2.33, 6.44 5.53 3.36, 9.10   
    Current 2.67 1.78, 4.02 2.32 1.53, 3.54 3.18 2.15, 4.74 4.46 3.03, 6.58 5.87 3.98, 8.68   
Pack-yearsd             
    0 1.00  1.13 0.73, 1.75 1.71 1.14, 2.57 1.79 1.20, 2.68 2.93 2.00, 4.29 0.12 0.06, 0.19 
    <10 (low) 2.40 1.51, 3.81 1.73 1.04, 2.89 3.30 2.11, 5.17 3.96 2.56, 6.13 4.95 3.20, 7.65   
    ≥10 (high) 3.34 2.21 5.06 3.17 2.07, 4.85 3.48 2.35, 5.23 4.54 3.03, 6.82 6.48 4.33, 9.72   

Abbreviations: CI, confidence interval; HR, hazard ratio; RERI, relative excess risk due to interaction.

a

Adjusted for pesticide use, working uncovered, smoking, body mass index, formal education, years of education, and follow-up interval.

b

Adjusted for fertilizer use, working uncovered, smoking, body mass index, formal education, years of education, and follow-up interval.

c

Adjusted for fertilizer use, pesticide use, smoking, body mass index, formal education, years of education, and follow-up interval.

d

Adjusted for body mass index, formal education, years of education, and follow-up interval.

Finally, we conducted analyses using daily arsenic intake. These results were qualitatively very similar to the results for well-water arsenic concentration and are therefore not shown here.

DISCUSSION

In the present large prospective cohort study, we evaluated on the additive scale the interaction between arsenic exposure and measures of sun exposure, fertilizer and pesticide use, and smoking status in men. We observed an apparent synergistic effect between smoking and arsenic exposure, as measured by well-water arsenic concentration, creatinine-adjusted urinary total arsenic concentration, and daily arsenic intake, and between fertilizer use and arsenic exposure, as measured by water arsenic concentration and daily arsenic intake only. Results of this study suggested that men in this cohort with a history of smoking and high fertilizer use may be more susceptible to incident skin lesions than are those with no smoking history or fertilizer use, even at the same level of arsenic exposure.

Arsenic exposure has been previously linked to increased risk of skin lesions based on findings from cross-sectional studies (2, 4, 6, 7, 11, 16, 18). Cross-sectional analysis of the HEALS baseline cohort data showed synergistic effects between the highest level of arsenic exposure (>113 μg/L) and tobacco smoking on risk of prevalent skin lesions in men. That study also observed possible synergistic effects between higher levels of arsenic exposure (28.1 μg/L–113.0 μg/L and 113.10–864.0 μg/L) and fertilizer use in men (RERI = 1.0, 95% confidence interval: −0.2, 2.2) (16). Prior analyses suggested that despite its necessary role in the development of skin lesions, arsenic exposure alone may not be sufficient to cause skin lesions. The present study reports, for the first time, findings from prospective analyses attempting to identify possible causal partners that modify the effect of arsenic on risk of skin lesions based on individual-level data.

Research suggests that an important limitation in understanding differential susceptibility to arsenic exposure can be attributed to the biologic complexities of arsenic metabolism (28). In humans, inorganic arsenic ingested through drinking water is taken up through red blood cells and distributed primarily to the liver, kidneys, spleen, lungs, and skin (29). It has been hypothesized that the methylation of arsenic may be important in this process (1, 3, 28–31). Decreased methylation capacity inhibits the conversion of inorganic arsenic to a form with lower tissue affinity and allows unmethylated arsenic to be deposited in hair, bone, and skin. The present study did not incorporate arsenic methylation measures directly. However, investigation of biologic cofactors known to be responsible for decreased methylation capacity, such as smoking, may help identify determinants of arsenic metabolism and thus elucidate the underlying mechanism of interindividual variability in risk of skin lesions (16, 31).

In addition to its effects on methylation, arsenic exposure may also adversely alter 1 or more DNA repair processes. Recent studies have shown that arsenic exposure decreases DNA repair capacity (32, 33). Thus, it is possible that tobacco-carcinogen–induced DNA damages would be less efficiently repaired in individuals exposed to arsenic through drinking water.

The results of the present study provide evidence for an interaction between arsenic exposure and smoking status on the additive scale. Our results suggest higher hazard ratios for individuals who are former smokers than for those who are current smokers. Separate analyses of pack-years yielded similar results. However, former smokers had greater average pack-years of exposure with greater variability (17 pack-years; standard deviation, 18.8) than did current smokers (15 pack-years; standard deviation, 13.3), which potentially accounts for these differences in association estimates.

Smoking has been well-characterized as a risk factor in several cancers, including cancers of the bladder, oral cavity, pancreas, and esophagus (20, 29); however, results of prior studies have not consistently established a significant association between smoking and nonmelanoma skin cancers (20, 34–36). Our findings suggest evidence for a significant synergistic effect between smoking and arsenic exposure on incident skin lesions, which may be potentially explained by the biologic consequences of decreased methylation due to smoking and may have considerable implications for the association between smoking and skin cancer risk. Additionally, our results suggest an increased risk of skin lesions even in individuals at the lowest level of arsenic exposure. Although theoretically, arsenic exposure is necessary to cause skin lesions, the results of the present study also suggest that perhaps there is a mechanism through which smoking alone can increase the risk of skin lesions.

The potential mechanisms through which exposure to fertilizers leads to increased risk of arsenic-induced skin lesions remain unknown. Reports have linked exposure to fertilizers and pesticides to occupational contact dermatitis (16, 37–39). We provide evidence of a notable synergistic interaction between water arsenic exposure and fertilizer use. Further investigation is necessary to determine the biologic basis of this interaction and whether exposure to fertilizers and pesticides can further impact the progression of arsenic-induced skin lesions to skin cancers.

The data collected from the HEALS cohort permit various assessments of arsenic exposure, allowing for a more thorough understanding of the adverse impact of arsenic, even at the lowest levels of exposure. Additionally, the study provides detailed information about exposure status and length of use for various occupational and lifestyle risk factors, including smoking and excess sun exposure. Unlike in prior studies, we were able to prospectively investigate effect modification of the relation between arsenic exposure and incident skin lesions. Moreover, we assessed interactions on the additive scale and found evidence of a synergistic effect in relation to both water- and urine-based arsenic exposure measures. Self-reported information was utilized regarding well-use history, sun exposure, smoking habits, and pesticide and fertilizer use. Although the validity of self-reported well use was good, one potential limitation of the study is misclassification of variables pertaining to occupational and environmental exposures, which could, in turn, have resulted in underestimation of the observed effects.

Our analysis is potentially limited by recall bias. Individuals were asked to report exposure to environmental risk factors such as pesticides and fertilizers, smoking, and working outdoors. It is likely that some participants misreported exposure to these factors. However, we do not suspect that individuals reported these exposures differentially based on their arsenic exposure. Therefore, the results reported here are most likely conservative estimates of the effects of these modifiers on the risk of skin lesions due to arsenic exposure. We attempted to control for potential confounding due to socioeconomic status by including variables regarding education level. Because of the prospective nature of the analysis, our results were likely not impacted by selection bias or reverse causality.

Ultimately, the identification of potential risk factors that are likely to impact skin lesion risk is useful in understanding differential susceptibility between populations. Our findings suggest that smoking status and exposure to certain occupational risk factors, such as fertilizer use, may increase synergistically with arsenic exposure to increase the risk of skin lesions in Bangladeshi men in this cohort. Because these skin lesions are often precursors to arsenic-induced skin cancers and are potentially markers of susceptibility to other health effects, the findings in the present study suggest that certain modifiable environmental factors may enhance the risk of skin cancer and other health outcomes (including internal cancers) due to arsenic exposure. With longer follow-up of the HEALS cohort, we will be able to examine this question empirically in the future.

Abbreviations

    Abbreviations
  • HEALS

    Health Effects of Arsenic Longitudinal Study

  • RERI

    relative excessive risk for the interaction

Author affiliations: Department of Health Studies, University of Chicago, Chicago, Illinois (Stephanie Melkonian, Maria Argos, Brandon L. Pierce, Paul J. Rathouz, Habibul Ahsan); Department of Environmental Medicine, New York University, New York, New York (Yu Chen); Columbia University and University of Chicago Research Office in Bangladesh, Mohakhali, Dhaka, Bangladesh (Tariqul Islam, Alauddin Ahmed, Emdadul H. Syed); Department of Environmental Health Sciences, Mailman School of Public Health, Columbia University, New York, New York (Faruque Parvez, Joseph Graziano); and Departments of Medicine and Human Genetics and Cancer Research Center, University of Chicago, Chicago, Illinois (Habibul Ahsan).

This work was supported by the National Institutes of Health (grants P42ES010349, R01CA102484, R01CA107431, and CA014599).

The authors thank all HEALS fieldworkers and staff for their ongoing commitment to the study.

Conflict of interest: none declared.

References

1.
Chen
Y
Factor-Litvak
P
Howe
GR
, et al.  . 
Arsenic exposure from drinking water, dietary intakes of B vitamins and folate, and risk of high blood pressure in Bangladesh: a population-based, cross-sectional study
Am J Epidemiol
 , 
2007
, vol. 
165
 
5
(pg. 
541
-
552
)
2.
Haque
R
Mazumder
DN
Samanta
S
, et al.  . 
Arsenic in drinking water and skin lesions: dose-response data from West Bengal, India
Epidemiology
 , 
2003
, vol. 
14
 
2
(pg. 
174
-
182
)
3.
Hopenhayn-Rich
C
Biggs
ML
Kalman
DA
, et al.  . 
Arsenic methylation patterns before and after changing from high to lower concentrations of arsenic in drinking water
Environ Health Perspect
 , 
1996
, vol. 
104
 
11
(pg. 
1200
-
1207
)
4.
Mazumder
DN
Haque
R
Ghosh
N
, et al.  . 
Arsenic in drinking water and the prevalence of respiratory effects in West Bengal, India
Int J Epidemiol
 , 
2000
, vol. 
29
 
6
(pg. 
1047
-
1052
)
5.
Parvez
F
Chen
Y
Argos
M
, et al.  . 
Prevalence of arsenic exposure from drinking water and awareness of its health risks in a Bangladeshi population: results from a large population-based study
Environ Health Perspect
 , 
2006
, vol. 
114
 
3
(pg. 
355
-
359
)
6.
Smith
AH
Arroyo
AP
Mazumder
DN
, et al.  . 
Arsenic-induced skin lesions among Atacameño people in Northern Chile despite good nutrition and centuries of exposure
Environ Health Perspect
 , 
2000
, vol. 
108
 
7
(pg. 
617
-
620
)
7.
Tondel
M
Rahman
M
Magnuson
A
, et al.  . 
The relationship of arsenic levels in drinking water and the prevalence rate of skin lesions in Bangladesh
Environ Health Perspect
 , 
1999
, vol. 
107
 
9
(pg. 
727
-
729
)
8.
Chen
CL
Hsu
LI
Chiou
HY
, et al.  . 
Ingested arsenic, cigarette smoking, and lung cancer risk: a follow-up study in arseniasis-endemic areas in Taiwan
JAMA
 , 
2004
, vol. 
292
 
24
(pg. 
2984
-
2990
)
9.
Hopenhayn-Rich
C
Biggs
ML
Fuchs
A
, et al.  . 
Bladder cancer mortality associated with arsenic in drinking water in Argentina
Epidemiology
 , 
1996
, vol. 
7
 
2
(pg. 
117
-
124
)
10.
Díaz-Barriga
F
Santos
MA
Mejía
JJ
, et al.  . 
Arsenic and cadmium exposure in children living near a smelter complex in San Luis Potosí, Mexico
Environ Res.
 , 
1993
, vol. 
62
 
2
(pg. 
242
-
250
)
11.
Ahsan
H
Chen
Y
Parvez
F
, et al.  . 
Health Effects of Arsenic Longitudinal Study (HEALS): description of a multidisciplinary epidemiologic investigation
J Expo Sci Environ Epidemiol
 , 
2006
, vol. 
16
 
2
(pg. 
191
-
205
)
12.
Ahsan
H
Perrin
M
Rahman
A
, et al.  . 
Associations between drinking water and urinary arsenic levels and skin lesions in Bangladesh
J Occup Environ Med
 , 
2000
, vol. 
42
 
12
(pg. 
1195
-
1201
)
13.
Argos
M
Parvez
F
Chen
Y
, et al.  . 
Socioeconomic status and risk for arsenic-related skin lesions in Bangladesh
Am J Public Health
 , 
2007
, vol. 
97
 
5
(pg. 
825
-
831
)
14.
World Health Organization
Arsenic in Drinking Water
 , 
2001
Geneva, Switzerland
World Health Organization
 
15.
Singer
JD
Willet
JB
It's about time: using discrete-time survival analysis to study duration and the timing of events
J Educ Stat
 , 
1993
, vol. 
18
 
2
(pg. 
155
-
195
)
16.
Chen
Y
Graziano
JH
Parvez
F
, et al.  . 
Modification of risk of arsenic-induced skin lesions by sunlight exposure, smoking, and occupational exposures in Bangladesh
Epidemiology
 , 
2006
, vol. 
17
 
4
(pg. 
459
-
467
)
17.
Xia
Y
Wade
TJ
Wu
K
, et al.  . 
Well water arsenic exposure, arsenic induced skin-lesions and self-reported morbidity in inner Mongolia
Int J Environ Res Public Health
 , 
2009
, vol. 
6
 
3
(pg. 
1010
-
1025
)
18.
Ahsan
H
Chen
Y
Parvez
F
, et al.  . 
Arsenic exposure from drinking water and risk of premalignant skin lesions in Bangladesh: baseline results from the Health Effects of Arsenic Longitudinal Study
Am J Epidemiol
 , 
2006
, vol. 
163
 
12
(pg. 
1138
-
1148
)
19.
Hertz-Picciotto
I
Smith
AH
Holtzman
D
, et al.  . 
Synergism between occupational arsenic exposure and smoking in the induction of lung cancer
Epidemiology
 , 
1992
, vol. 
3
 
1
(pg. 
23
-
31
)
20.
De Hertog
SA
Wensveen
CA
Bastiaens
MT
, et al.  . 
Relation between smoking and skin cancer
J Clin Oncol
 , 
2001
, vol. 
19
 
1
(pg. 
231
-
238
)
21.
Nermell
B
Lindberg
AL
Rahman
M
, et al.  . 
Urinary arsenic concentration adjustment factors and malnutrition
Environ Res.
 , 
2008
, vol. 
106
 
2
(pg. 
212
-
218
)
22.
Chizmar
JF
A discrete-time hazard analysis of the role of gender in persistence in the economics major
J Econ Educ
 , 
2000
, vol. 
31
 
2
(pg. 
107
-
118
)
23.
Twisk
JW
Longitudinal data analysis. A comparison between generalized estimating equations and random coefficient analysis
Eur J Epidemiol
 , 
2004
, vol. 
19
 
8
(pg. 
769
-
776
)
24.
Zeger
SL
Liang
KY
Longitudinal data analysis for discrete and continuous outcomes
Biometrics
 , 
1986
, vol. 
42
 
1
(pg. 
121
-
130
)
25.
Li
R
Chambless
L
Test for additive interaction in proportional hazards models
Ann Epidemiol
 , 
2007
, vol. 
17
 
3
(pg. 
227
-
236
)
26.
Knol
MJ
van der Tweel
I
Grobbee
DE
, et al.  . 
Estimating interaction on an additive scale between continuous determinants in a logistic regression model
Int J Epidemiol
 , 
2007
, vol. 
36
 
5
(pg. 
1111
-
1118
)
27.
Hosmer
DW
Lemeshow
S
Confidence interval estimation of interaction
Epidemiology
 , 
1992
, vol. 
3
 
5
(pg. 
452
-
456
)
28.
Loffredo
CA
Aposhian
HV
Cebrian
ME
, et al.  . 
Variability in human metabolism of arsenic
Environ Res.
 , 
2003
, vol. 
92
 
2
(pg. 
85
-
91
)
29.
Chen
YC
Su
HJ
Guo
YL
, et al.  . 
Interaction between environmental tobacco smoke and arsenic methylation ability on the risk of bladder cancer
Cancer Causes Control
 , 
2005
, vol. 
16
 
2
(pg. 
75
-
81
)
30.
Gamble
MV
Liu
X
Ahsan
H
, et al.  . 
Folate, homocysteine, and arsenic metabolism in arsenic-exposed individuals in Bangladesh
Environ Health Perspect
 , 
2005
, vol. 
113
 
12
(pg. 
1683
-
1688
)
31.
Hopenhayn-Rich
C
Biggs
ML
Smith
AH
, et al.  . 
Methylation study of a population environmentally exposed to arsenic in drinking water
Environ Health Perspect
 , 
1996
, vol. 
104
 
6
(pg. 
620
-
628
)
32.
Andrew
AS
Burgess
JL
Meza
MM
, et al.  . 
Arsenic exposure is associated with decreased DNA repair in vitro and in individuals exposed to drinking water arsenic
Environ Health Perspect
 , 
2006
, vol. 
114
 
8
(pg. 
1193
-
1198
)
33.
Banerjee
M
Sarma
N
Biswas
R
, et al.  . 
DNA repair deficiency leads to susceptibility to develop arsenic-induced premalignant skin lesions
Int J Cancer
 , 
2008
, vol. 
123
 
2
(pg. 
283
-
287
)
34.
Freedman
DM
Sigurdson
A
Doody
MM
, et al.  . 
Risk of basal cell carcinoma in relation to alcohol intake and smoking
Cancer Epidemiol Biomarkers Prev
 , 
2003
, vol. 
12
 
12
(pg. 
1540
-
1543
)
35.
Hunter
DJ
Colditz
GA
Stampfer
MJ
, et al.  . 
Risk factors for basal cell carcinoma in a prospective cohort of women
Ann Epidemiol
 , 
1990
, vol. 
1
 
1
(pg. 
13
-
23
)
36.
van Dam
RM
Huang
Z
Rimm
EB
, et al.  . 
Risk factors for basal cell carcinoma of the skin in men: results from the Health Professionals Follow-up study
Am J Epidemiol
 , 
1999
, vol. 
150
 
5
(pg. 
459
-
468
)
37.
Lazarov
A
Yair
M
Lael
E
, et al.  . 
Airborne irritant contact dermatitis from phosphates in a fertilizer factory
Contact Dermatitis
 , 
2002
, vol. 
46
 
1
(pg. 
53
-
54
)
38.
Lensen
G
Jungbauer
F
Gonçalo
M
, et al.  . 
Airborne irritant contact dermatitis and conjunctivitis after occupational exposure to chlorothalonil in textiles
Contact Dermatitis
 , 
2007
, vol. 
57
 
3
(pg. 
181
-
186
)
39.
Pecegueiro
M
Contact dermatitis due to nickel in fertilizers
Contact Dermatitis
 , 
1990
, vol. 
22
 
2
(pg. 
114
-
115
)