Residential proximity to hazardous waste sites and risk of end-stage renal disease.

Introduction

End-stage renal disease (ESRD) is characterized by total renal failure making hemodialysis or organ transplantation necessary. Some ESRD can be explained by underlying disease processes such as diabetes mellitus or polycystic kidney disease (1). The etiologic origin of many of the remaining ESRD cases has not been determined and might be of environmental origin. Environmental exposures that have been linked to chronic renal failure include exposures to hydrocarbons and metals.

Occupational lead exposures in battery plant workers, lead production or smelter workers, and workers involved in sandblasting of surfaces painted with lead (2-6) have been associated with ESRD. High-level lead exposures in childhood (7,8) and consumption of "moonshine" (presumably contaminated with lead) have been linked to ESRD as well (9). Lead can have a direct effect on the kidneys or an indirect effect through increased blood pressure. Higher body burdens of lead have been found in patients with chronic renal failure than in control patients (10,11).

Cadmium reaches the highest body concentrations in the renal cortex and has been associated with renal dysfunction (12). Human studies were based on occupational exposures and environmental exposures in Japan (Itai-Itai disease), Belgium, and New Zealand. The primary findings associated with chronic cadmium exposures in animals as well as humans were proteinuria, glucosuria, aminoaciduria, and progressive reduction in the glomerular filtration rate which was identified by the significantly increased serum creatinine and beta 2-microglobulin concentrations. It was found that tubular damage was not reversible, and could progress (12,14).

Table 1

The Specific or Contributing Diagnoses for the End-Stage Renal
Disease Cases, as Reported by HCFA(*) End-Stage Renal Disease
Study,
New York, 1994

Diagnosis                     Number             Percent

Glomerulonephritis              88                  55
Hypertension                    53                  33
Urinary system disease          12                   7
Metabolic disease                3                   2
Connective tissue disease        2                   1
Blood disease                    2                   1
Renal failure, unspecified       1       [less than] 1

* This information was available for 161 of the 216 cases
interviewed
Table 2

Demographic Characteristics of Cases and Controls End-Stage Renal
Disease Study, New York, 1994

Variable                           No. of Cases      No. of
Controls

Education

High school or less                    115                   93
More than high school                   98                  123
Don't know                               3                   -

Income (annual)

[less than or equal to] 15,000          41                   27
[greater than] 15,000                  134                  175
Refused/Unknown                         41                   14

Ever smoked

Yes                                    141                  122
No                                      75                   94

Ever drank alcohol

Yes                                    179                  172
No                                      37                   44

Case reports and analytic studies suggest potential associations between work-related hydrocarbon or solvent exposures and ESRD (9,15-18). Agents in this category include solvents used as cleaning agents or degreasers,gasoline, diesel and jet fuel, paints, mineral turpentine, and glue. Compared to other causes of ESRD, solvent exposures have been specifically linked to an increased incidence of glomerulonephritis (19,20).

Heavy metals and solvents, as well as other toxic agents, are often present at hazardous waste sites. They pose potential exposures for residents near such sites through contaminated food, air, and water, and direct contact with surface water and soil. The relationships between these exposures and ESRD have not been investigated.

Materials and Methods

Study Population

This case-control study was conducted in 20 counties of New York State. These counties excluded New York City and included urban, suburban, and rural areas. The counties contain 70% of the sites listed on the New York Inactive Hazardous Waste Site Registry. Sites in the registry are sites on the federal National Priorities List and additional sites designated under New York's program.

All newly diagnosed cases of ESRD reported to the Health Care Financing Administration (HCFA) from January 1, 1992 through December 31, 1993, who resided in one of the 20 New York counties were eligible for participation in the study. Since the 1972 amendments to the Social Security Act extended medical coverage to ESRD patients of all ages, almost all cases are covered by Medicare regardless of socioeconomic status and therefore reported to HCFA. Cases with diabetes; polycystic kidney disease; lupus; congenital nephropathy; and nephropathy due to malignancy, analgesic abuse, or infections were not eligible for further study and, therefore, were excluded. For a large number of cases, no diagnosis was specified in the HCFA files. Attempts were made to contact these cases and determine their eligibility according to the exclusion criteria listed above. The age range for participation in the study, as determined from the date of birth on the HCFA database, was 18 to 60 years.

Eligible cases were mailed a letter introducing the study and inviting participation. This was followed by telephone contact to enlist those willing to participate. Telephone numbers of cases were obtained by searching credit databases for the case name and address, or the case address only. For cases for whom telephone numbers were not available, the following procedures were used in sequence: (1) directory assistance was contacted; (2) a follow-up letter was sent asking the individual to contact the principal investigator; and (3) a certified letter was sent asking the individual to contact the principal investigator.

Controls were selected through random-digit dialing within the 20 selected counties of the state of New York (21). Controls were pair-matched to cases by age ([+ or -]5 years), race, and sex. Controls reporting a history of any type of kidney disease were excluded.

Data Collection

A standardized questionnaire was administered to all participants by telephone. The questionnaire asked about residential histories, occupations, hobbies, health, and lifestyle behaviors. Interviewers were aware of the case-control status of participants but not their exposure status.

[TABULAR DATA FOR TABLE 3 OMITTED]

Exposure Assessment

The basic method of exposure assessment has previously been described (22). In this study, residential addresses of participants were classified on the basis of proximity to hazardous waste sites and likelihood of exposure. Proximity to sites was calculated using the latitudes and longitudes of the full residential addresses. This was determined by matching the actual addresses to 1990 Bureau of the Census TIGER files and by comparing the addresses with the location of the center of each site. To determine the likelihood of exposure, a polygon with 25 sectors within a 1-mile radius was established from the center of each site. For each sector, the likelihood of residential exposure to solvents and metals was determined by reviewing site histories, sampling data, and reports on other investigations conducted at the site. A series of standard worksheets provided the basis for the evaluation. The likelihood of exposure (low, medium, high, or unknown) of a polygon sector was then assigned to any residence within that particular sector. For solvents, a residence was classified as having a likelihood of exposure greater than low if there was evidence of residential exposure by off-site contamination via ambient air vapor, private wells, or groundwater (basement) vapor for any period before diagnosis. For metals, a residence was classified as having a likelihood of exposure greater than low if there was evidence of residential exposure by off-site contamination via air particulates or private well contamination. Up to five residential addresses per participant were assessed with this method. To account for the varying lengths of residential histories available, analyses were first conducted for all participants and then for those who had lived in the 20 New York counties for at least 10 years.

Data Analyses

Matched pair analyses were conducted by testing differences in the discordant pairs with the McNemar's chi-square test (23). Discordant pair odds ratios (ORs) and 95 percent confidence intervals (CIs) were calculated to determine the magnitude of the associations of interest and the precision of the OR estimates. Descriptive and matched pairs analyses were conducted using SAS (24).

Conditional logistic regression modeling was conducted to test the effects of interest, assess additional covariates (retained if p[less than]0.05), and adjust for potential confounders (25). The regression analyses were conducted using the EGRET (26) statistical package.

Results

Of 378 cases selected from the HCFA files, by the time of contact 18 were deceased, 9 were too ill to be interviewed, 3 were mentally impaired, 2 did not speak English or Spanish, and 1 could not be contacted in jail. Of the remaining 345 cases, 209 were successfully interviewed. Fifty-one cases refused to be interviewed upon telephone contact. No telephone contact was possible with 85 cases who could not be traced. Another 50 interviews were completed from an additional 316 cases who did not have a diagnosis specified on the HCFA files, but who were found to be eligible upon screening. Therefore, a total of 259 cases were interviewed. An equal number of controls were interviewed, for a total of 518 completed interviews.

For those cases for whom information was available from the HCFA files, a comparison was made by sex, race, and age between the participants who completed interviews and those who did not. No significant differences were observed between the two groups for these variables.

The specific or contributing diagnoses for 161 of the 216 cases interviewed were known from the HCFA data (Table 1). All participants were also asked about their disease history, and 43 participant pairs were removed from the data because the respondents reported having physician-diagnosed diabetes or lupus. The final number of pairs remaining for analysis was 216.

There were more men (134, 62 percent) than women (82, 38 percent) among the cases. The racial distribution was 71 percent white and 28 percent black. One pair was matched on Hispanic and one on Asian origin. The distribution of the demographic characteristics of income, education, and smoking and drinking behaviors of the cases and controls is shown in Table 2. More cases than controls had smoked 100 cigarettes or more in their lifetime and had no more than a high school education.

A total of 1,522 addresses of the 432 participants were in New York State. For 399 (26 percent) of the addresses, the exposure status was not determined. Of the 1,123 addresses that could be matched, 479 (43 percent) were within a 1-mile radius of a hazardous waste site and 644 (57 percent) were not. The exposure status of the unmatched addresses could not be determined because either they were not in the 20 counties for which exposure assessment was conducted or because the address information was incomplete. The number of addresses that could be matched was about the same for cases (72 percent) and controls (75 percent). Because cases were diagnosed with ESRD in 1992 and 1993, only addresses where participants had lived before 1992 were considered in the analyses.

A total of 259 participants had ever lived within a 1-mile radius of a hazardous waste site in the 20 New York counties. Most of these participants had lived in the 0.6- to 1.0-mile [TABULAR DATA FOR TABLE 4 OMITTED] vicinity of a site (27 percent) or the 0.3- to 0.6-mile vicinity (23 percent); only 10 percent of the participants had lived within 0.3 mile of a site. The remaining 173 participants either had never lived close to a site in the 20 counties or their addresses could not be matched with the Census files. The median number of years participants lived within a 1-mile radius of a site was 12 years; the range was from less than 1 year to 58 years.

Participants had lived within a 1-mile radius of any of 317 hazardous waste sites. Specific information on solvent and metal contamination was known for 63 sites: 52 contained solvents only, 1 contained metals only, and 10 sites contained both. The types of metals found at the sites were lead, mercury, cadmium, and chromium. Some participants lived in the 1-mile radii of two or more sites. The median number of sites participants were exposed to was 2 (range: from 1 to 16 sites).

Thirty participants had a high (2) or medium (28) probability of exposure to contaminants from sites. All of the exposures were to solvents; 1 of the 30 also had been exposed to metals. The median number of years of exposure was 6.5 (range: less than 1 year to 29 years).

Elevated ORs were found for the associations between ESRD and residential proximity to a site, the number of years lived in proximity to a site, and a medium or high probability of exposure (Table 3). However, the associations were not statistically significant at p [less than or equal to] 0.05. In the conditional logistic regression analyses only the exposure variable "distance" (ever lived within a 1-mile radius of a site) was retained. The "years of exposure" variable (number of years lived in proximity to a site) was not included in further modeling because the models did not converge when both "distance" and "years" were included in the model due to collinearity.

The variables shown in Table 4 were individually assessed as potential confounders of the relationship between ESRD and "distance." Only the variables "high blood pressure" and "gout" were found to produce a difference of more than 10 percent between the OR adjusted for age, sex, and race by matching and the ORs adjusted for the matched variables and hypertension or gout. "Kidney disease in family," a history of kidney disease in first degree relatives, was retained in the model as an additional independent risk factor with a p value [less than or equal to] 0.05.

The final model is shown in Table 5, with an OR for "ever lived in 1-mile radius" of 1.63 (95 percent CI: 0.87-3.03). Strong relationships were found for high blood pressure, gout, and a family history of kidney disease. The wide CIs reflect the lack of precision obtained with a small number of pairs.

Additional analyses were conducted to examine the influence on the findings of the individuals' potential for exposure. An analysis restricted to those participants (74 pairs) who had lived in the 20 New York counties for at least 10 years also yielded an elevated, but statistically not significant, OR of 1.5 (p = 0.31, 95 percent CI: 0.69-3.28) for having ever lived within a 1-mile radius.

Discussion

A statistically not significant elevated odds ratio was found for having ever lived within a 1-mile radius of a hazardous waste site and ESRD. The variable that could have refined the exposure definition, the higher probability of exposure based on site assessments, did not show a statistically significant association with ESRD. From the literature, heavy, metal and solvent exposures were identified as risk factors for ESRD. In this study, it was not possible to determine the effects of metals and solvents because of the lack of specific information for all sites. Those individuals who were found to have a higher likelihood of exposure, based on [TABULAR DATA FOR TABLE 5 OMITTED] the site assessments, had been exposed primarily to solvents. The failure to find a statistically significant association with ESRD for these solvent-exposed people might have been due to the low number of people exposed or to the absence of an effect.

Residential vicinity to hazardous waste sites was used as a surrogate for measurements of heavy metals and solvents in the water, soil, or air. However, living near hazardous waste sites might also be associated with other social or economic factors. Education and income were considered potential confounders since people with lower socioeconomic status might be more likely to live close to hazardous waste sites. Neither factor was found to confound the association between living within a 1-mile radius to a site and ESRD, and they did not enter the model as independent risk factors.

A potential bias was introduced by the limited lifetime exposure information available concerning residences near hazardous waste sites. Participants were asked to provide only their current and previous four addresses. Further, only addresses within the 20 counties studied could be assessed for proximity to and exposure from sites. Because only newly diagnosed cases were enrolled in this study, it seems unlikely that ESRD influenced residential histories. Having kidney disease before complete renal failure might have contributed to more stable residential histories. Yet, restricting the analyses to those who had lived for 10 or more years in any of the 20 New York counties also yielded an elevated, statistically not significant, OR. Information on exposures from polluting industries active at the time of the study was not available.

The size of the sample was another limitation of this study, expressed in the lack of precision for the point estimates by wide CIs. The study design of pair matching made it necessary to eliminate some pairs from the analyses because of reported exclusion criteria such as diabetes. Persons with diabetes were excluded based on the assumption that diabetes has a greater effect in causing ESRD than low-level environmental exposures. However, environmental exposures might contribute to the diabetic predisposition to ESRD, and this issue needs to be considered in future studies.

To avoid biased subject selection, all cases meeting the study's case definition were invited to participate. A small percentage of ESRD cases who were not recipients of Medicare benefits, and therefore not reported to HCFA, might have been missed. Selection bias caused by a refusal to participate and the inability to contact cases might have posed a potential problem. However, based on the information provided by HCFA, no significant differences were observed in the sex, race, and age distributions between those who completed interviews and those who did not: Controls were randomly selected from the general population; however, the refusal rate among controls could not be assessed with the methods used in this study.

Of the variables considered potential confounders, only hypertension and gout were retained in the final model. Both gout (27) and hypertension (28,29) are known risk factors for ESRD and, in some studies (30,31), have been linked to lead exposure. In this study, it was not possible, however, to determine whether participants developed hypertension before or after the onset of ESRD. Familial predisposition to kidney disease had also been previously reported (32); however, it has not been determined whether this predisposition might be exacerbated by exposures to environmental contaminants.

This was the first study that attempted to assess the potential association between exposures from hazardous waste sites and ESRD. However, no statistically significant association was found and some of the study's limitations may have influenced the results. Additional studies are therefore needed to determine whether living near hazardous waste sites increases the risk of renal disease and whether chemicals or other factors explain this association.

REFERENCES

1. Health Care Financing Administration (1990), HealthCare Financing Research Report - End Stage Renal Disease, 1988, HCFA, Baltimore, Md.

2. Schirmer, J. (1990), "Lead Risks Overlooked in Sandblasters?" (Letter to the Editor), Am.J. Pub. Health, 80(10):1275.

3. Cooper, W.C. (1988), "Deaths from Chronic Renal Disease in U.S. Battery and Lead Production Workers," Envtl. Health Persp., 78:61-3.

4. Selevan, S.G., J.H. Jones, P.J. Landrigan, and F.B. Stern (1985), "Mortality of Lead Smelter Workers," Am. J. Epidemiol., 122(4):673-83.

5. McMichael, A.J., and H.M. Johnson (1982), "Long-Term Mortality Profile of Heavily-Exposed Lead Smelter Workers," J. Occup. Med., 24(5):375-8.

6. Malcolm, D., and H.A.R. Barnett (1982), "A Mortality Study of Lead Workers 1925-76," Br. J. Ind. Med., 39(4):110-404.

7. Craswell, P.W., J. Price, P.D. Boyle, V.J. Heazlewood, H. Baddeley, H.M. Lloyd, B.J. Thomas, B.W. Thomas, and G.M. Williams (1986), "Chronic Lead Nephropathy in Queensland: Alternative Methods of Diagnosis," Austl. N.Z.J. Med., 16(1):11-19.

8. Osterloh, J.D., C.E. Becker, B. Behrens, B.P. Bernard, D.J. Menke, J.D. Ordonez, J.V. Selby, and E. Tepper (1989), "Body Burdens of Lead in Hypertensive Nephropathy," Arch. Envtl. Health, 44(5):304-10.

9. Steenland, N.K., C.W. Ferguson, F.D. Port, and M.J. Thun (1990), "Occupational and Other Exposures Associated with Male End-Stage Renal Disease: A case-control study," Am. J. Pub. Health, 80(2):153-7.

10. Koster, J., A. Erhardt, C. Mohl, E. Ritz, and M. Stoeppler (1989), "Mobilizable Lead in Patients with Chronic Renal Failure," Eur. J. Clin. Invest., 19(2):228-33.

11. Ritz, E., J. Mann, and A. Wiecek (1988), "Does Lead Play a Role in the Development of Renal Insufficiency?" Contrib. Nephrol., 64:43-8.

12. Agency for Toxic Substances and Disease Registry (1989), Toxicological Profile for Cadmium., ATSDR/TP-88/08, U.S. Dept. of Health and Human Services, Public Health Service, Atlanta, Ga.

13. Kjellstrom, T. (1986), "Renal Effects,"Cadmium and Health: A toxicological and epidemiological appraisal, Vol. II, ed. L. Friberg, C.G. Elinder, T. Kjellstrom, and G.F. Nordber, CRC Press, Inc., Boca Raton, Fl., pp. 21-109.

14. Roels, H.A., A.M. Bernard, J.P. Buchet, R.R. Lauwerys, A. Vos, and M. Oversteyns (1989), "Health Significance of Cadmium Induced Renal Dysfunction: A five year follow up," Br. J. Ind. Med., 46(11):755-64.

15. Daniell, W.E., W.G. Couser, and L. Rosenstock (1988), "Occupational Solvent Exposure and Glomerulonephritis," JAMA, 259(15):2280-3.

16. Churchill, D.N., A. Fine, and M.H. Gault (1983), "Association Between Hydrocarbon Exposure and Glomerulonephritis - An Appraisal of the Evidence," Nephron., 33(3):169-72.

17. Lauwerys, R., A. Bernard, J.P. Buchet, and C. Viau (1985), "Kidney Disorders and Hepatotoxicity from Organic Solvent Exposure," Scand. J. Work. Envtl. Health, 11(Suppl 1):83-90.

18. Venkataraman, G. (1981), "Renal Damage and Glue Sniffing," BMJ, 283(6304):1467.

19. Zimmerman S.W., G.J. Beirne, and K. Groehler (1975), "Hydrocarbon Exposure and Chronic Glomerulonephritis," Lancet, 2(7927):199-201.

20. Lagrue, G. (1976), "Hydrocarbon Exposure and Chronic Glomerulonephritis," Lancet, 1(7970):1191.

21. Survey Sampling, Inc. (1991), "Statistical Characteristics of Random Digit Telephone Samples," SS-Information Sheet., Survey Sampling, Inc., Fairfield, Ct., pp. 1-4.

22. Marshall, E.G., M,R. Cayo, N.S. Geary, and P.A. Lauridsen (1993), "Residential Exposure for a Reproductive Health Study of Multiple Hazardous Waste Sites," J. Exposure Anal. Envtl. Epidemiol, 3(Suppl. 1):87-98.

23. Breslow, N.E., and N.E. Day (1990), "Statistical Methods in Cancer Research: Volume 1 - The analysis of case-control studies, International Agency for Research on Cancer, Lyon, France.

24. SAS Institute Inc. (1989), SAS Language and Procedures, Version 6, 1st ed., SAS Institute Inc., Cary, NC.

25. Kelsey, J.L., A.S. Evans, and W.D. Thompson (1986), Methods in Observational Epidemiology, Oxford University Press, NY.

26. Statistical and Epidemiology Research Corporation (SERC) (1991), EGRET, Revision 2, Seattle, Wa.

27. Patial, R.K., and V.K. Sehgal (1992), "Non-Oliguric Acute Renal Failure in Gout," Indian J. Med. Sci., 46(7):201-4.

28. Lopes, A.A., K. Hornbuckle, S.A. James, and F.K. Port (1994), "The Joint Effects of Race and Age on the Risk of End-Stage Renal Disease Attributed to Hypertension," Am. J. Kidney Dis., 24(4):554-60.

29. Innes, A., R.P. Burden, A.M. Davidson, P.A. Johnston, and A.G. Morgan (1993), "Clinical Features of Benign Hypertensive Nephrosclerosis at Time of Renal Biopsy," Q.J. Med., 86(4):271-5.

30. Craswell, P.W., P.D. Boyle, and J. Price (1987), "Patterns of Lead Excretion in Patients with Gout and Chronic Renal Failure - A Comparative German and Australian study," The Science of the Total Environment, 66:17-28.

31. Staessen, J.A., A. Amery, C.J. Bulpitt, R. Fugard, R.R. Lauwerys, P. Lijnen, H. Roels, and L. Thijs (1994), "Is a Positive Association Between Lead Exposure and Blood Pressure Supported by Animal Experiments?" Curr. Opin. Nephrol. Hypertens., 3(3):257-63.

32. Freedman, B.I., V.M. Buckalew, B.J. Spray, and A.B. Tuttle (1993), "The Familial Risk of End-Stage Renal Disease in African Americans," Am. J. Kidney Dis., 21(4):387-93.

Corresponding Author: H. Irene Hall, Ph.D., Epidemiologist, Centers for Disease Control and Prevention, NCCDPHP/DCPC/ESB, Mailstop K-55, 4770 Buford Hwy. N.E., Atlanta, GA 30341.