Socioeconomic Status and Risk for Arsenic-Related Skin Lesions in Bangladesh

Abstract

Objectives. Arsenic contamination of groundwater is a severe public health crisis in Bangladesh, where the population is exposed to arsenic in drinking water through tube wells used for groundwater collection. In this study, we explored the association between socioeconomic status and arsenic toxicity.

Methods. We used baseline data from 11438 men and women who were recruited into the Health Effects of Arsenic Longitudinal Study (HEALS), a prospective cohort study on the health effects of arsenic exposure in Bangladesh. We conducted analyses with logistic regression and generalized estimating equations.

Results. We found a strong dose–response association with all measures of arsenic exposure and skin lesions. We also found that the effect of arsenic was modified by land ownership on a multiplicative scale, with an increased risk among non–land owners associated with well water arsenic (P=.04) and urinary total arsenic concentrations (P=.03).

Conclusions. Our study provides insight into potentially modifiable host characteristics and identifies factors that may effectively target susceptible population subgroups for appropriate interventions.

Beginning in the 1970s, hand-pumped tube wells were installed in Bangladesh to provide a source of pathogen-free groundwater for consumption. Since then, the number of tube wells within the country has exponentially increased, and these wells are the population’s primary source of drinking water. Unfortunately, the groundwater in Bangladesh has been naturally contaminated with high levels of arsenic, a phenomenon discovered only after decades of exposure and an epidemic of arsenical skin lesions.1^,2

Many of the human health effects of arsenic have been proven, and arsenic has been classified as a human carcinogen.3 Epidemiologic evidence has shown a significant association between the consumption of arsenic through drinking water and cancers of the skin, lung, bladder, liver, and kidney.4^–8 Premalignant skin lesions are an early manifestation and hallmark of arsenic toxicity and may indicate increased future risk for arsenic-related cancer.9 However, there is evidence of marked interindividual variability in arsenic-induced premalignant skin lesions and cancer.9 It has been suggested that differential susceptibility to arsenic-induced skin lesions may be attributed to host characteristics10^,11; therefore, because of the magnitude of the arsenic problem in Bangladesh, a thorough examination of potentially modifiable host characteristics is warranted.

Socioeconomic status (SES) has been identified as an important determinant of health across a broad range of health issues.12 SES is a multidimensional construct that includes education, occupation, income, wealth, and residence. However, the measure of this construct in less-developed countries creates a challenge because adequate data may not be available. Therefore, the examination of several measures of SES is advantageous because it increases the likelihood that some dimension of the SES construct is captured. No previous studies of arsenic exposure have explicitly examined effect modification by SES in arsenic toxicity. We sought to determine whether selected SES indicator variables modify the association between various measures of arsenic exposure and premalignant skin lesions among 11438 men and women in Araihazar, Bangladesh.

METHODS

In 2000, the Health Effects of Arsenic Longitudinal Study (HEALS)—an ongoing population-based study of both the short-term and long-term health effects of arsenic exposure—was launched in Araihazar, Bangladesh. The selection of cohort participants, study design, and methods have been described in detail elsewhere and are briefly summarized here.13

Participants

The cohort participants are residents of Araihazar, Bangladesh—a well-defined 25-square-kilometer rural area east of the capital city of Dhaka. All tube wells in the area (n=5966) were first tested for arsenic concentration as part of a precohort survey. A complete enumeration of the study area population (n=65876) was conducted as a part of this survey. The preliminary well survey findings and details of the study methods used have been published elsewhere.14 From the enumeration, a list of eligible married couples was generated, and regular users of tube wells were selected for recruitment into the study. Individuals were eligible for participation if they were married, lived in the bari (a cluster of household dwellings occupied by members of the extended family) for at least 3 years, and were aged 18 to 75 years. The rationale for these selection criteria has been published elsewhere.13

The field team recruited couples in the order they were recorded on the precohort survey list. Once an individual was located (i.e., matched by name and age and by husband’s name for female respondents), eligibility was verified. Eligible individuals and their spouses were recruited into the study. If the spouse was not present during the initial visit, the recruiter returned to the bari on another day to recruit the spouse. The recruiter continued to enroll eligible individuals and their spouses until a maximum of 3 couples from each tube well was obtained; this restriction was adopted to reduce the correlation among individuals in the cohort. Between October 2000 and May 2002, 11746 men and women were recruited; this included 4801 married couples and 2144 participants whose spouses were not enrolled in the cohort. There was a 97.5% participation rate among those who were approached for recruitment. Comparison with the precohort survey showed that our cohort population was a representative sample of the study population in terms of arsenic and geographic distributions (data not shown).

Questionnaire

A structured questionnaire composed in Bengali was administered to participants by trained interviewers. The questionnaire assessed sociodemographic characteristics, current and past tube well use, typical water consumption patterns, food consumption frequency of 39 items common to the population, occupational exposures, and smoking habits. The semiquantitative food consumption frequency questionnaire was developed for this target population and validated.15 The interviewers were blind to the tube well water arsenic concentrations for the study region.

Clinical Examination

During the baseline interview, each participant was examined by a trained physician for the presence of premalignant skin lesions in accordance with a structured protocol.13 The physicians were specially trained for the detection and diagnosis of these skin lesions. Premalignant skin lesions were categorized as melanosis, leucomelanosis, and keratosis.16 The presence, location, and extent of each skin lesion type were documented. Physicians were blind to the participants’ well water arsenic concentration, and no histopathologic confirmation of these lesions was conducted.

Prevalent cases of premalignant skin lesions (n=714) were ascertained during the baseline survey. Melanosis was characterized by the hyperpigmentation of the skin over wide body surface areas, leucomelanosis was characterized by the hypopigmentation of the skin over wide body surface areas, and keratosis was characterized by the general thickening of the skin on palms and soles.16 Among the 714 skin lesion cases, 421 (59.0%) had melanosis only; 194 (27.2%) had melanosis and keratosis; 44 (6.2%) had leucomelanosis and keratosis; 29 (4.1%) had leucomelanosis only; 14 (2.0%) had melanosis, leucomelanosis, and keratosis; and 12 (1.7%) had melanosis and leucomelanosis. We dichotomized skin lesion status into presence or absence of any skin lesions.

Exposure Assessment

Three measures of individual-level arsenic exposure were assessed: well water arsenic concentration, creatinine-adjusted urinary total arsenic concentration, and cumulative arsenic exposure (CAE). Well water was analyzed for arsenic concentration with graphite furnace atomic absorption spectrometry that has a detection limit of 5 μg/L.17 Water samples that had less than 5 μg/L were subsequently reanalyzed with inductively coupled plasma-mass spectrometry that has a detection limit of 0.1 μg/L.18 Urinary total arsenic concentration was analyzed with graphite furnace atomic absorption spectrometry in accordance with the method used by Nixon et al.19 Urinary creatinine was measured with a colorimetric Sigma Diagnostics Kit (Sigma Diagnostics, St Louis, Mo), and urinary total arsenic concentration was subsequently expressed as micrograms per gram creatinine. CAE was calculated by multiplying well water arsenic concentration times the estimated amount of water consumed per year times the number of years the tube well had been used for drinking.13 A constant arsenic concentration in each tube well was assumed for calculation of the CAE for the duration of use by each participant. This assumption was previously established for tube wells in our study area.14 Furthermore, individuals who reported either use of another tube well before their current well or use of a secondary tube well in addition to their current well had those additional sources of exposure included in their CAE calculations.

SES Variables

Education, land ownership, television ownership, weekly household cooking oil consumption, and a food index were considered to be indicators of SES.20 The baseline questionnaire measured years of education, land ownership, television ownership, and food frequency for all participants; weekly household cooking oil consumption was asked only of female respondents. The amount reported by the female participant was adjusted for total household size and also was used for her spouse. The food index was created by adding the total number of days per week that each participant reported eating fish, beef or lamb, poultry, eggs, or dried beans. These food items were selected as indicators of household wealth on the basis of their correlation with the other selected SES indicators. Household cooking oil consumption and the food index were selected as indicators of SES independent of their potential contribution to nutritional status. All continuous SES variables were dichotomized at their median for our analyses. The correlation among these SES indicator variables was weak (Pearson correlation coefficients 0.22–0.35), which suggests that they each captured different components of SES variability.

Statistical Analysis

Of the 11746 individuals enrolled at baseline, 308 individuals (<3%) were eliminated from our analysis. Reasons for exclusion included having declined a clinical examination (n=210), having skin lesions unrelated to arsenic (n=96), and having not reported age (n=2). Thus, the sample eligible for our cross-sectional analysis was 11438. Additionally, we excluded from all regression analyses those participants who were missing data on at least 1 SES indicator or other covariates (n=353). Individuals who were missing arsenic exposure data were excluded only from the analysis of the missing exposure measure—323 individuals for urinary total arsenic concentration data and 381 for CAE data. Well water data were not missing for any participant. Our logistic regression analyses used the generalized estimating equation method and included 11085 individuals (681 with skin lesions and 10404 without lesions) for well water arsenic concentration, 10771 individuals (666 with skin lesions and 10105 without lesions) for urinary total arsenic concentration, and 10719 individuals (668 with skin lesions and 10051 without lesions) for CAE.

We used linear regression to evaluate the association between SES indicators after we adjusted for gender, age, and body mass index (kg/m²). This was done separately for both well water arsenic concentration and creatinine-adjusted urinary total arsenic concentration. For descriptive purposes, we used 2-sided χ² tests to compare subjects by skin lesion status with respect to gender, age, and SES indicators. We used logistic regression to estimate prevalence odds ratios (ORs) and 95% confidence intervals (CIs) for the association between these characteristics and skin lesion status after we adjusted for gender, age, and well water arsenic concentration. Participants were then categorized into quintiles of well water arsenic, urinary total arsenic concentration, and CAE separately in accordance with the distribution among the total cohort eligible for analysis (n=11438). The quintile containing the lowest level for each measure was the reference category.

We defined premalignant skin lesions—the outcome of interest in this analysis—as the presence of any type of arsenic-related skin lesion (melanosis, leucomelanosis, or keratosis). Prevalence ORs and their 95% CIs were estimated with logistic regression models. Because several individuals drank water from the same tube well, we used the generalized estimating equation method21 to calculate effect estimates and their CIs, which accounted for the correlated exposure data. Although different measures of relative risk can be estimated from data of cross-sectional studies, we deemed prevalence ORs an appropriate measure of effect for this study.22 Prevalence odds ratios were adjusted for all SES indicator variables, gender, age (continuous), body mass index (continuous), smoking status (current or past cigarette smoker vs never smoked), and occupation (indicators for laborers, unemployed, and homemakers vs business).

Multiplicative interaction was assessed by evaluating the Wald statistic for the cross-product interaction term from the regression model and was interpreted as the statistical probability for interaction. A statistical probability of less than 0.05 for the interaction term suggested that the slope of the 2 dose–response trends was statistically different when stratified by the effect modifier. All analyses were performed with SAS version 9.0 (SAS Institute Inc, Cary, NC).

RESULTS

The baseline clinical evaluation identified 714 premalignant skin lesion cases. The average baseline well water arsenic concentration was 165.8 μg/L among arsenical skin lesion cases and 97.2 μg/L among nonlesion controls, the average baseline creatinine-adjusted urinary total arsenic concentration was 426.6 μg/g creatinine among arsenical skin lesion cases and 271.5 μg/g creatinine among non-lesion controls, and the average baseline CAE was 2213.2 mg among arsenical skin lesion cases and 984.1 mg among nonlesion controls. Results from the linear regression models showed television ownership to be the only SES indicator consistently associated with arsenic exposure, with individuals who did not own a television having higher well water and urinary total arsenic concentrations (data not shown) than television owners.

Table 1 ▶ shows the distribution of cohort participants by skin lesion status with respect to gender, age, and SES indicators. Men and individuals aged 35 years and older had the highest prevalence of skin lesions. We also saw a significant difference in skin lesion prevalence by SES indicator strata, with the highest skin lesion prevalence consistently among the categories indicating lower SES. The association between SES and premalignant skin lesions was examined with logistic regression, and we saw significant associations between all SES indicators and skin lesions after we adjusted for gender, age, and well water arsenic concentration. An increased risk for skin lesions was consistently found among the lower-SES categories.

TABLE 1—

Baseline Skin Lesion Status Among 11 438 Adults in the HEALS Cohort: Araihazar, Bangladesh, 2000–2002

	Skin Lesion Status			Age-Adjusted Skin Lesion Prevalence, %
Sample Characteristic	Present (n = 714), No. (%)	Absent (n = 10 724), No. (%)	Pb	Men	Women	PORa (95% CIa)
Gender			0.001
Women	130 (18.2)	6432 (60.0)		NA	2.27	1.0 (NA)
Men	584 (81.8)	4292 (40.0)		9.96	NA	4.8 (3.9, 5.9)
Age, y			0.001
18–35	141 (19.7)	5418 (50.5)		3.68	0.74	1.0 (NA)
36–75	573 (80.3)	5306 (49.5)		9.34	2.34	2.7 (2.2, 3.3)
Education, y			0.001
> 2	290 (40.6)	5391 (50.3)		7.96	1.62	1.0 (NA)
≤ 2	424 (59.4)	5327 (49.7)		12.40	2.73	1.6 (1.3, 1.8)
Missing		6 (0.1)
Owns a television			0.001
Yes	175 (24.5)	3780 (35.2)		6.79	1.64	1.0 (NA)
No	537 (75.2)	6944 (64.8)		11.60	2.61	1.7 (1.4, 2.0)
Missing	2 (0.3)
Owns land			0.091
Yes	331 (46.4)	5321 (49.6)		7.88	2.19	1.0 (NA)
No	382 (53.5)	5387 (50.2)		11.82	2.30	1.3 (1.1, 1.5)
Missing	1 (0.1)	16 (0.2)
Cooking oil usec (mL/week)			0.001
> 125	258 (36.1)	5172 (48.2)		8.22	1.69	1.0 (NA)
≤ 125	430 (60.2)	5320 (49.6)		12.09	2.79	1.5 (1.3, 1.8)
Missing	26 (3.6)	232 (2.2)
Food indexd (days)			0.001
> 6	312 (43.7)	5533 (51.6)		8.06	1.71	1.0 (NA)
≤ 6	402 (56.3)	5191 (48.4)		12.36	2.77	1.6 (1.3, 1.8)

Note. HEALS = Health Effects of Arsenic Longitudinal Study, POR = prevalence odds ratio, CI = confidence interval; NA = not applicable.

^aAdjusted for gender, age, and well water arsenic concentration.

^bTwo-sided χ² test; missing values were excluded from the statistical comparison.

^cAdjusted for household size.

^dThe food index was created by adding the total number of days per week that each participant reported eating fish, beef or lamb, poultry, eggs, or dried beans. These food items were indicators of household wealth on the basis of their correlation with the other selected socioeconomic indicators.

A strong dose–response association was found between all measures of arsenic exposure and premalignant skin lesions in our cross-sectional analysis. Table 2 ▶ shows adjusted prevalence ORs and their 95% CIs for the associations between well water arsenic exposure and premalignant skin lesions stratified by SES indicators. We observed significant effect modification on a multiplicative scale for the association between premalignant skin lesions and well water arsenic concentration by land ownership, with a higher risk among non–land owners (P = .04). Table 3 ▶ shows the adjusted prevalence ORs and 95% CIs for the associations between urinary total arsenic concentration and premalignant skin lesions stratified by SES indicators. We found significant effect modification by land ownership (P = .03) for the effects of urinary total arsenic concentrations. Table 4 ▶ shows similar dose–response trends for the associations between CAE and premalignant skin lesions; however, there were no interactions with SES indicators.

TABLE 2—

Prevalence Odds Ratios (95% Confidence Intervals) for Premalignant Skin Lesions Associated With Well Water Arsenic Concentration by Socioeconomic Status Indicators Among 11 438 Adults in the HEALS Cohort: Araihazar, Bangladesh, 2000–2002

	Well Water Arsenic Concentration, μg/La
Sample Characteristic	< 7	7–38	39–90	91–177	> 177	Interaction P
Education, y						0.51
≤ 2	1.0	2.01 (1.14, 3.53)	3.62 (2.11, 6.21)	3.77 (2.20, 6.45)	5.72 (3.39, 9.64)
> 2	1.0	1.82 (1.04, 3.19)	2.22 (1.29, 3.82)	2.82 (1.67, 4.77)	4.42 (2.65, 7.36)
Owns a television						0.95
No	1.0	2.07 (1.26, 3.40)	2.83 (1.75, 4.59)	3.36 (2.10, 5.39)	5.27 (3.32, 8.37)
Yes	1.0	1.53 (0.78, 3.02)	3.10 (1.67, 5.75)	3.15 (1.69, 5.87)	4.56 (2.44, 8.52)
Owns land						0.04
No	1.0	1.78 (0.99, 3.21)	3.17 (1.80, 5.57)	4.47 (2.59, 7.74)	6.04 (3.53, 10.34)
Yes	1.0	2.08 (1.23, 3.51)	2.72 (1.65, 4.49)	2.39 (1.43, 3.99)	4.30 (2.64, 7.01)
Cooking oil,b mL/week						0.93
≤ 125	1.0	2.07 (1.18, 3.65)	3.98 (2.32, 6.84)	3.66 (2.13, 6.29)	5.55 (3.29, 9.37)
> 125	1.0	1.75 (1.00, 3.05)	1.84 (1.06, 3.17)	2.91 (1.74, 4.86)	4.57 (2.76, 7.57)
Food indexc, days						0.99
≤ 6	1.0	1.98 (1.12, 3.51)	2.90 (1.66, 5.07)	3.18 (1.83, 5.51)	5.15 (3.04, 8.72)
> 6	1.0	1.82 (1.06, 3.15)	2.88 (1.72, 4.83)	3.36 (2.02, 5.59)	4.85 (2.95, 7.96)

Note. HEALS = Health Effects of Arsenic Longitudinal Study. Prevalence odds ratios were estimated with the generalized estimating question and were adjusted for gender, age, body mass index, smoking status, occupation, and all other variables presented in the table.

^aQuintiles of exposure; quintile 1, < 7 μg/L, was the reference group.

^bAdjusted for household size.

^c The food index was created by adding the total number of days per week that each participant reported eating fish, beef or lamb, poultry, eggs, or dried beans. These food items were indicators of household wealth on the basis of their correlation with the other selected socioeconomic indicators.

TABLE 3—

Prevalence Odds Ratios (95% Confidence Intervals) for Premalignant Skin Lesions Associated With Creatinine-Adjusted Urinary Total Arsenic Concentration by Socioeconomic Status Indicators Among 11 438 Adults in the HEALS Cohort: Araihazar, Bangladesh, 2000–2002

	Creatinine-Adjusted Urinary Total Arsenic Concentration, μg/g Cra
Sample Characteristic	≤ 35	36–66	67–114	115–204	> 204	Interaction P
Education, y						0.93
≤ 2	1.0	1.75 (1.09, 2.79)	2.04 (1.29, 3.21)	2.48 (1.56, 3.93)	5.16 (3.31, 8.03)
> 2	1.0	1.75 (1.07, 2.86)	2.42 (1.52, 3.87)	3.55 (2.19, 5.75)	5.02 (3.10, 8.13)
Owns a television						0.09
No	1.0	1.55 (1.04, 2.30)	1.96 (1.32, 2.91)	2.41 (1.62, 3.58)	4.31 (2.94, 6.32)
Yes	1.0	2.21 (1.19, 4.09)	2.70 (1.53, 4.78)	4.32 (2.39, 7.82)	7.66 (4.22, 13.90)
Owns land						0.03
No	1.0	1.38 (0.85, 2.23)	2.13 (1.35, 3.36)	2.58 (1.60, 4.15)	5.92 (3.79, 9.26)
Yes	1.0	2.14 (1.33, 3.43)	2.27 (1.44, 3.58)	3.28 (2.06, 5.22)	4.13 (2.59, 6.60)
Cooking oil use,b mL/week						0.63
≤ 125	1.0	1.63 (1.00, 2.64)	2.44 (1.55, 3.86)	2.68 (1.67, 4.30)	5.40 (3.43, 8.50)
> 125	1.0	1.91 (1.19, 3.06)	1.86 (1.16, 3.00)	3.29 (2.05, 5.28)	4.70 (2.94, 7.50)
Food index,c days						0.18
≤ 6	1.0	1.53 (0.93, 2.51)	2.08 (1.29, 3.35)	2.42 (1.48, 3.94)	4.27 (2.70, 6.76)
> 6	1.0	2.02 (1.26, 3.22)	2.38 (1.50, 3.77)	3.58 (2.25, 5.69)	6.48 (4.09, 10.25)

Note. HEALS = Health Effects of Arsenic Longitudinal Study. Prevalence odds ratios were estimated with the generalized estimating equation and were adjusted for gender, age, body mass index, smoking status, occupation, and all variables presented in the table.

^aQuintiles of exposure; quintile 1, ≤ 35 μg/g Cr, was the reference group.

^bAdjusted for household size.

^c The food index was created by adding the total number of days per week that each participant reported eating fish, beef or lamb, poultry, eggs, or dried beans. These food items were indicators of household wealth on the basis of their correlation with the other selected socioeconomic indicators.

TABLE 4—

Prevalence Odds Ratios (95% Confidence Intervals) for Premalignant Skin Lesions Associated With Cumulative Arsenic Exposure by Socioeconomic Status Indicators Among 11 438 Adults in the HEALS Cohort: Araihazar, Bangladesh, 2000–2002

	Cumulative Arsenic Exposure, mg a
Sample Characteristic	< 62	62–224	225–583	584–1490	> 1490	Interaction P
Education, y						0.54
≤ 2	1.0	1.89 (1.14, 3.15)	3.08 (1.85, 5.12)	3.90 (2.36, 6.44)	5.41 (3.30, 8.88)
> 2	1.0	1.55 (0.88, 2.72)	1.92 (1.10, 3.36)	3.09 (1.81, 5.26)	5.29 (3.18, 8.79)
Owns a television						0.79
No	1.0	1.86 (1.16, 2.98)	2.64 (1.65, 4.23)	3.63 (2.30, 5.72)	5.49 (3.51, 8.58)
Yes	1.0	1.42 (0.73, 2.76)	2.27 (1.22, 4.24)	3.29 (1.79, 6.04)	5.14 (2.81, 9.39)
Owns land						0.22
No	1.0	1.63 (0.95, 2.79)	3.11 (1.84, 5.26)	3.80 (2.27, 6.36)	6.27 (3.79, 10.36)
Yes	1.0	1.87 (1.11, 3.15)	2.06 (1.22, 3.49)	3.33 (2.01, 5.51)	4.59 (2.81, 7.50)
Cooking oil use,b mL/week						0.38
≤ 125	1.0	1.50 (0.89, 2.53)	2.66 (1.61, 4.39)	3.80 (2.32, 6.20)	5.48 (3.38, 8.88)
> 125	1.0	2.11 (1.25, 3.56)	2.33 (1.34, 4.02)	3.08 (1.82, 5.22)	5.20 (3.15, 8.57)
Food index,c days						0.91
≤ 6	1.0	1.54 (0.89, 2.66)	2.60 (1.52, 4.45)	3.50 (2.07, 5.92)	5.08 (3.04, 8.50)
> 6	1.0	1.99 (1.18, 3.35)	2.46 (1.46, 4.14)	3.51 (2.11, 5.85)	5.78 (3.54, 9.42)

Note. HEALS = Health Effects of Arsenic Longitudinal Study. Prevalence odds ratios were estimated with the generalized estimating equation and were adjusted for gender, age, body mass index, smoking status, occupation, and all variables presented in the table.

^aQuintiles of exposure; quintile 1, < 62 μg, was the reference group.

^bAdjusted for household size.

^c The food index was created by adding the total number of days per week that each participant reported eating fish, beef or lamb, poultry, eggs, or dried beans. These food items were indicators of household wealth on the basis of their correlation with the other selected socioeconomic indicators.

DISCUSSION

In this cross-sectional analysis of baseline data from a large cohort study, we observed borderline significant effect modification of the association between premalignant skin lesions and arsenic exposure by land ownership. Our population-based study was the first, to our knowledge, to show this heterogeneity of effect of arsenic on skin lesion risk by SES. Because of the relative homogeneity of our study population, these findings are rather striking. We used several indicators of SES that are intercorrelated; however, each variable explains a different component of SES variability in this population. We found that land ownership appears to be the best indicator of SES for this rural population with regard to the health effects of arsenic.

The food index was evaluated for bias that may have arisen from the potential misclassification of vegetarian participants as having low SES. Only a small proportion of our study cohort appeared to be strictly vegetarian (n=709), which was defined as no consumption of meat, fish, or eggs. In an analysis that excluded individuals who were identified as strict vegetarians, we evaluated effect modification of the associations between arsenic exposure measures and arsenical skin lesions by the food index, and we found that the effect estimates and overall risk trends appeared to be very similar to those in the overall study population (data not shown). Thus, we believe no significant bias of the food index was introduced with the inclusion of vegetarian participants in our analyses.

Our dose–response findings are consistent with previous studies that examined the association between arsenic exposure and premalignant skin lesions.23^,24 There is an underlying biological rationale for also exploring differential susceptibility to arsenic-induced skin lesions. Molecular epidemiologic evidence suggests that there is marked variability in arsenic metabolism among exposed individuals.25 Factors that include gender, age, smoking status, nutrition, and genetic polymorphisms may influence arsenic metabolism.26 Because of the extent of arsenic exposure within the Bangladeshi population and the public health crisis that has resulted from this exposure, our study focused on modifiable characteristics and can be used to readily identify subpopulations at greater risk for the health effects of arsenic exposure.

Recent studies by Hadi and Parveen27 and Sikder et al.28 found SES to be a significant risk factor for prevalent skin lesions, although no adjustment for the level of arsenic exposure was made in either analysis. In our study, we had a sufficiently large sample for evaluating effect modification by SES, and we found a significant interaction between arsenic exposure and land ownership with respect to premalignant skin lesion risk.

The use of prevalent cases of premalignant skin lesions is unlikely to have biased our findings. Our assumption is based on the fact that mortality from arsenic-induced lesions is not immediate; therefore, we did not expect survivor bias to be present in this analysis. We also did not expect prevalent cases to be sufficiently different from incident cases of skin lesions with regard to the modifying effect of SES on the arsenic-induced skin lesion association.

Modification of the association between arsenic exposure and skin lesions by SES may be because of differential ascertainment of exposure status or skin lesions by SES. Skin lesions were assessed through a physical examination by trained physicians who followed a structured protocol. Furthermore, both interviewers and physicians were blind to the well water arsenic concentration of participants. It is unlikely that laboratory analyses of well water and urinary total arsenic concentrations would yield systematically different exposure measures by SES. Therefore, we believe it is improbable that an ascertainment bias by SES could have distorted the findings of this study.

Epidemiologic research of other exposures, diseases, and populations has frequently shown health disparity and variations in disease prevalence by SES. Therefore, our finding of increased risk for arsenic-induced skin lesions among lower-SES individuals was not an unexpected one. The consistency of this finding across arsenic exposure measures (well water arsenic and urinary total arsenic concentrations) by land ownership also is not surprising. However, contrary to our previous belief, effect modification was not seen with CAE. This arsenic exposure measure is a function of both dose and duration. Our findings suggest that interaction between SES and arsenic exposure may be associated with a mechanism for which dose of exposure is important. The CAE measure is problematic in this regard because of the fact that an individual with a high-dose exposure of short duration would have the same cumulative exposure as an individual with a low-dose exposure of long duration. Thus, evaluating effect modification with CAE may attenuate any modifying effect.

Although the etiology of the interaction between SES and arsenic is presently not understood and needs to be explored further, it is an important factor for a public health intervention in this population. Our study population is an agrarian society. Land-based agriculture provides livelihood for the majority of the rural population, and land is a meaningful measure of SES. By evaluating land ownership, we have identified a higher-risk sub-population among arsenic-exposed individuals. Future population-based interventions that are focused on the prevention of arsenic-induced lesions should be designed with the specific needs and characteristics of this higher-risk population in mind.

The assessment of arsenical skin lesions in this study enhanced our knowledge about individuals who are susceptible to arsenic. Skin lesions are much more common than skin cancers among individuals who are exposed to arsenic from drinking water.29 Unlike skin cancer, which has a latency of decades, skin lesions appear within a few years of arsenic exposure. Epidemiologic evidence suggests that skin lesions may be associated with increased risk for arsenic-induced cancers.30^,31 Therefore, skin lesions may be considered an intermediate endpoint for arsenic-induced cancers. Knowledge or characteristics of susceptible individuals who may also be at a higher risk for arsenic-induced cancers is critical for intervention and prevention of future disease burden in this population.

Future studies of susceptibility to arsenic-induced disease should explore the mechanistic pathways by which SES modifies the association between arsenic exposure and premalignant skin lesions. Potential mechanisms include nutrition, health behaviors and awareness, and access to and use of medical care. For example, it has been suggested that nutritional status may affect one’s ability to methylate arsenic, which may be a factor in the differential susceptibility to arsenic-induced disease.25 Future studies also should examine other host characteristics associated with differential susceptibility to arsenic-induced disease.

Arsenic exposure through drinking water among the Bangladeshi population is an important public health concern. Yet, there has been little research on modifiable risk factors for arsenic toxicity. The results of our cross-sectional study show that SES—specifically land ownership—modifies the association between arsenic exposure and premalignant skin lesions, with a higher risk among the lower-SES group. Further research is needed to uncover the biological pathways that may be responsible for this heterogeneity of effect and to exploit that knowledge for developing and evaluating effective interventions.