Original Investigation Incident ESRD Among Participants in a Lead Surveillance Program Ritam Chowdhury, PhD,1 Lyndsey Darrow, PhD,1 William McClellan, MD, PhD,1 Stefanie Sarnat, PhD,2 and Kyle Steenland, PhD1,2 Background: Very high levels of lead can cause kidney failure; data about renal effects at lower levels are limited. Study Design: Cohort study, external (vs US population) and internal (by exposure level) comparisons. Settings & Participants: 58,307 men in an occupational surveillance system in 11 US states. Predictor: Blood lead levels. Outcome: Incident end-stage renal disease determined by matching the cohort with the US Renal Data System (n 5 302). Measurements: Blood lead categories were 0-,5, 5-,25, 25-,40, 40-51, and .51 mg/dL, defined by highest blood lead test result. One analysis for those with data for race (31% of cohort) and another for the whole cohort after imputing race. Results: Median follow-up was 12 years. Among those with race information, the end-stage renal disease standardized incidence ratio (SIR; US population as referent) was 1.08 (95% CI, 0.89-1.31) overall. The SIR in the highest blood lead category was 1.47 (95% CI, 0.98-2.11), increasing to 1.56 (95% CI, 1.02-2.29) for those followed up for 5 or more years. For the entire cohort (including those with race imputed), the overall SIR was 0.92 (95% CI, 0.82-1.03), increasing to 1.36 (95% CI, 0.99-1.73) in the highest blood lead category (SIR of 1.43 [95% CI, 1.01-1.85] in those with $5 years’ follow-up). In internal analyses by Cox regression, rate ratios for those with 5 or more years’ follow-up in the entire cohort were 1.0 (0-,5 and 5-,25 mg/dL categories combined) and 0.92, 1.08, and 1.96 for the 25-,40, 40-51, and .51 mg/dL categories, respectively (P for trend 5 0.003). The effect of lead was strongest in nonwhites. Limitations: Lack of detailed work history, reliance on only a few blood lead tests per person to estimate level of exposure, lack of clinical data at time of exposure. Conclusions: Data suggest that current US occupational limits on blood lead levels may need to be strengthened to avoid kidney disease. Am J Kidney Dis. -(-):---. ª 2014 by the National Kidney Foundation, Inc. INDEX WORDS: End-stage renal disease (ESRD); lead; kidney disease; occupational exposure; Adult Blood Lead Epidemiology and Surveillance (ABLES).
L
ead exposure in the United States has declined since the 1970s.1 Blood lead levels in the United States decreased from 12.2 to 2.8 mg/dL from 1976 to 19912 and then to 1.3 mg/dL in 2008.3 However, some US adults still have high occupational exposure to lead. The National Institute for Occupational Safety and Health (NIOSH) estimated in the 1980s that more than 3 million workers in the United States were exposed.4 NIOSH currently conducts an adult blood lead surveillance program that captures those with higher exposure; data from 37 states indicated that approximately 130,000 adults had been tested for blood lead in 2005.5 The current US Occupational Safety and Health Administration (OSHA) standard calls for workers to be removed from exposure when they have: (1) a blood lead level of 50 mg/dL (construction workers) or (2) 60 mg/dL on a single test or an average of $50 mg/dL on a series of recent tests (other workers) and to not return until their blood lead level decreases to ,40 mg/dL.6 Twelve percent of US adults have diagnosed chronic kidney disease, with increasing numbers each year; 6% have stage 3 disease or higher Am J Kidney Dis. 2014;-(-):---
(glomerular filtration rate , 60 mL/min/1.73 m2).7 There currently are about 600,000 prevalent cases of end-stage renal disease (ESRD) in the United States.8 Very high lead exposure historically has been shown to lead to an increased risk of acute kidney injury.9 Chronic high levels of exposure resulting in blood lead levels . 70-80 mg/dL have been associated with lead nephropathy, characterized by tubular damage, in studies10 of both humans and animals, although some researchers are unconvinced by this evidence.11 Much of the evidence for humans for chronic kidney disease From the Departments of 1Epidemiology and 2Environmental Health, Rollins School of Public Health, Emory University, Atlanta, GA. Received September 1, 2013. Accepted in revised form December 4, 2013. Address correspondence to Kyle Steenland, PhD, Department of Epidemiology, Rollins School of Public Health, Emory University, 1518 Clifton Rd, Atlanta, GA 30322. E-mail: nsteenl@ emory.edu 2014 by the National Kidney Foundation, Inc. 0272-6386/$36.00 http://dx.doi.org/10.1053/j.ajkd.2013.12.005 1
Chowdhury et al
in relation to high lead exposures comes from occupational studies.12-15 In recent years, there also has been increasing evidence of lead’s negative effect on kidney function (eg, increased serum creatinine levels) at lower levels of lead exposure than studies in earlier occupational cohorts.10,16-26 Here, we have studied the incidence of ESRD in men tested for blood lead in 11 US states as part of the NIOSH-sponsored Adult Blood Lead Epidemiology and Surveillance (ABLES) program.
METHODS Data Sources and Study Participants The ABLES program started collecting state-level data for blood lead exposure in 1987.27 Initially, some states gathered data only for individuals with blood lead levels . 25 mg/dL, but subsequently began to collect data at lower levels. Blood lead tests were performed primarily because of occupational exposure. ABLES coverage increased from 4 states in 1987 to 41 states in 2012.27 NIOSH has collected data regarding industry for a limited number of ABLES individuals (n 5 6,999) with blood lead levels $ 25 mg/dL.28 Of these, 62% were in manufacturing (43% in battery production or renovation), 10% were in construction, 7% were in metal mining, and 21% were in other industries or data were unavailable. Only 4% of these individuals were exposed nonoccupationally (eg, by lead in shooting ranges). We obtained data from 11 state ABLES programs— Connecticut, California, Ohio, Minnesota, Iowa, Pennsylvania, New York, New Jersey, Wisconsin, Michigan, and Massachusetts— from their year of first participation through 2008. These states made up 4%, 18%, 10%, 2%, 3%, 3%, 29%, 11%, 6%, 3%, and 11%, respectively, of the overall cohort. These states were chosen because they had the most individuals with blood lead data and had data that went back the farthest in time. Furthermore, adding additional states would not have been possible due to the increased costs and our budget constraints. We excluded everyone first tested after 2005 to avoid short follow-up. We excluded those missing information for date of birth, test date, or blood lead levels. We categorized each blood lead level reading into 1 of 5 categories: 0-,5, 5-,25, 25-,40, 40-51, and .51 mg/dL. Categories of ,25, 25-40, and .40 mg/dL traditionally have been used to categorize occupational blood lead levels, whereas the lowest category we used (,5 mg/dL) essentially is equivalent to nonoccupational US blood lead levels. We divided the upper category (.40 mg/dL) into 2 groups with an approximately equal number of ESRD cases to explore whether there was excess risk confined to those with particularly high blood lead levels. For individuals with more than 1 blood lead test, blood lead category was defined as the highest category ever achieved. Among those eligible for inclusion, we first selected everyone who had ever had a blood lead level $ 25 mg/dL (categories 3-5). We then randomly selected an equal number of people from categories 1 and 2 (50% from each category), stratified by state. Three states (Massachusetts, Michigan, and Wisconsin) opted to do their own data processing and matching with the National Death Index (NDI) and the US Renal Data System (USRDS). They followed the same selection pattern and sent us de-identified data. We restricted our analytic cohort to men because women represented only 15% of the cohort and only 9% of ESRD cases and were concentrated more heavily in the lowest blood lead category (40% had levels , 5 mg/dL compared to 11% for men), 2
suggesting that they may have included a higher percentage of nonoccupational blood leads, for example, pregnant women tested for lead during pregnancy (Susan Payne, California ABLES/ Occupational Lead Poisoning Prevention Program, personal communication, May 2013). We further excluded all people who were tested for the first time after the age of 70 years or before the age of 18 years and any implausible blood test results (.250 mg/dL). We excluded some ESRD cases (n 5 61) because their outcomes had occurred (ESRD diagnosis) before or on the same day as their first blood lead measurement. After the mentioned exclusions, we had a final analytic data set with 58,307 individuals.
Matching With NDI (deaths) and USRDS (incident ESRD) Matching with the NDI was done to determine vital status. We used first name, last name, date of birth, sex, race (when available), and social security number (when available) to match with the NDI through 2010 (2008 for Massachusetts and 2009 for Michigan and Wisconsin, all of which did their own matching). To determine a true match among possible multiple NDI matches, we selected those assigned a status code of 1 by NDI, indicating a high probability of a match. Past validation studies indicate that 95% of patients with ESRD are in the USRDS29; good records are kept, given the high cost of treatment, which is covered by Medicare or the US Department of Veterans Affairs. We used the same matching variables as we did for NDI for USRDS matching through 2010 (2008 for Michigan and Wisconsin). We accepted all matches so designated by the USRDS. Similar matching of other occupational cohorts has been done in the past.30-32
Adjustment for Missing Race The majority of our cohort was missing data for race (69%). Nonwhites in the United States have much higher ESRD rates than whites. Age- and sex-adjusted ESRD incidence rates (per 100,000) for African Americans and whites in 2010 in the United States were 92.4 and 27.5, respectively.33 Although race was not associated with lead level among those with known race (Table 1) and would not be expected to affect internal analyses, misclassified race could result in incorrect expected ESRD cases in external analyses (observed cases are unaffected because we had data for race for all of them). We chose 2 strategies to deal with the problem of missing race. First, we conducted an analysis restricted to the 31% with known race. Second, we imputed race for those missing race (although not for the 302 ESRD cases in the total cohort, for whom race was known) by multiple imputation (n 5 5) and ran a life table for each imputation. Details can be found in Item S1 (provided as online supplementary material). Briefly, we built a model that predicted race among those with data for race and used that model to predict race for those without data for race. The model correctly predicted race for w70% of individuals; given that race is known for 31% of the cohort, we can expect that race will be misclassified for w20% of the full cohort using imputation. Given that whites will be misclassified more than nonwhites, this remaining misclassification likely would cause some overestimation of expected ESRD cases in the full cohort analysis, resulting in conservative underestimates of the true rate ratios for exposed versus nonexposed (see Item S1 for details).
Analyses The NIOSH Life Table Analysis System was used to calculate person-years at risk and rates of ESRD incidence for the cohort and then compare these rates with those of the US population by standardized incidence ratios (SIRs), adjusted for age, race, sex, and calendar year.34,35 Am J Kidney Dis. 2014;-(-):---
Lead Exposure and Incident ESRD Table 1. Description of Demographics by Lead Category Highest Lead Category Achieved Total
0-,5 mg/dL
5-,25 mg/dL
25-,40 mg/dL
40-51 mg/dL
.51 mg/dL
58,307
6,832 (12%)
18,618 (32%)
21,440 (37%)
7,348 (12%)
4,069 (7%)
12.2
6.4
10.2
14.3
17
17.7
39.0 6 11.7
40.7 6 12.2
40.0 6 11.7
37.9 6 11.4
38.0 6 11.4
38.9 6 11.7
15,936 (27.3%) 2,315 (4.0%) 40,056 (68.7%) 0.13%
1,499 (22%) 222 (3%) 5,111 (75%) 0.13%
2,566 (14%) 416 (2%) 15,636 (84%) 0.14%
7,015 (32.7%) 954 (4.4%) 13,471 (62.8%) 0.12%
2,981 (4.06%) 423 (5.8%) 3,944 (53.7%) 0.12%
1,875 (46.1%) 300 (7.4%) 1,894 (46.6%) 0.14%
No. with single tests
28,540 (48.9%)
6,108 (89.4%)
12,710 (68.3%)
7,783 (36.3%)
1,445 (19.7%)
494 (12.1%)
Median no. of testsb
4
2
3
4
6
7
25.9 6 17.9
2.5 6 1.1
13.0 6 5.5
30.8 6 4.2
44.7 6 3.3
65.3 6 18.7
No. of participants Median follow-up (y) Mean age at first testa (y) Race White Nonwhite Missing/unknown % nonwhite among those with known race
Mean highest blood lead level (mg/dL)a No. with SSN data
15,233 (26.1%)
611 (8.9%)
2,079 (11.2%)
7,661 (35.7%)
3,104 (42.2%)
1,778 (43.7%)
Median year of birth
1959
1962
1961
1959
1956
1954
Median year of ESRD
2004
2007
2004
2004
2003
2003
302 (0.5%)
29 (0.4%)
79 (0.4%)
98 (0.5%)
44 (0.6%)
52 (1.2%)
No. with ESRD
Abbreviations: ESRD, end-stage renal disease; SSN, social security number. a Expressed as mean 6 standard deviation. b In those with more than 1 test.
Incidence rates of ESRD in the United States are available since 1973 and presently go through 2007. We used these US ESRD rates, extrapolated until 2010 (the end of our follow-up), as the referent rates to calculated expected ESRD cases, adjusted for age, race, sex, and calendar year. We calculated SIRs for all blood lead categories (1-5) and all categories combined, as well as by time since first exposure (0-5 or $5 years). Person-time at risk began at the first blood test. All person-years were assigned to the highest blood level ever attained regardless of the number of tests. For example, if a man had 3 tests with blood levels of 35, 43, and 48 mg/dL, he was assigned to the category 40-51 mg/dL, and all his person-time (beginning at his first blood test) was assigned to that category. Most people (68%) did not change blood lead category, whereas 24% changed category by only one category. Only 8% of the cohort changed blood lead category by more than one blood lead level. Person-time ended for everyone in 2010, the end of our NDI and USRDS follow-up, except for Massachusetts (2009), Michigan (2008), and Wisconsin (2008). In internal analyses, we calculated rate ratios using Poisson models (log person-years offset, correction for overdispersion) adjusted for sex, race, age, and calendar period (SAS, version 9.3; SAS Institute Inc). We compared lead categories 3-5 (25-,40, 40-51, and .51 mg/dL) to a reference category (,25 mg/dL). We combined categories 1 and 2 (0-,5 and 5-,25 mg/dL) due to the small number of cases in each of them. These analyses were based on the person-years calculated by the life table analyses and took into account the censoring occurring with mortality. We also conducted a trend test for ESRD incidence by assigning median values for each blood lead category in the regression, such that lead level became a continuous rather than a categorical variable. The test for trend was based on the significance of that variable. Finally, we also conducted internal analyses using Cox regression. In these analyses, we constructed 100-person risk sets for each case, randomly selected from those who attained the same age or older as their case’s age at diagnosis and restricted to those Am J Kidney Dis. 2014;-(-):---
who were born 6 2.5 years from the date of birth of their respective cases.
RESULTS Table 1 provides descriptive information about the cohort. There were 58,307 men. Forty-nine percent had only one blood test, whereas the rest had a median of 4 tests. Median follow-up was 12 years (increasing from 6.4 years in the lowest blood lead category to 17.7 years in the highest). There were 3,337 deaths in the cohort. The percentage of nonwhites was similar across blood lead categories, among those with known race. We did not have data regarding occupation or industry for the entire cohort. However, we had data for 72% of our California cohort (w7,600 men) for whether a blood test was occupational or nonoccupational (eg, from exposure to lead at a shooting range or lead paint in a residence). Of these, only 2% were nonoccupational. In the highest blood lead category, 82% had information on occupation, compared to 48% in the lowest category. Table 2 shows a comparison of incident ESRD in the cohort by lead category using the US population as referent (external comparison). Overall, there was a slight excess of ESRD incidence (SIR, 1.08; 95% confidence interval [CI], 0.89-1.31; 108 cases) compared to the general population. There was some elevation in ESRD incidence in the highest blood lead category (SIR, 1.47; 95% CI, 0.98-2.11), which was 3
(1.02-2.29)a (1.14-2.55)a (0.76-2.33) (1.28-4.32)a 1.56 1.74 1.39 2.48 26 26 14 12 Note: n 5 18,251. Abbreviations: CI, confidence interval; ESRD, end-stage renal disease; F/U, follow-up; SIR, standardized incidence ratio. a Statistically significant at P # 0.05. b There was 51% of the cohort with more than 1 blood test. Men with more than 1 blood test are likely to exclude any nonoccupational cases.
(0.70-1.73) (0.72-1.93) (0.40-1.65) (0.94-3.91) 1.13 1.22 0.87 2.06 21 18 9 9 (0.48-1.11) (0.60-1.49) (0.49-1.49) (0.49-2.49) 0.74 0.98 0.89 1.21 24 21 14 7 (0.11-1.62) (0.01-2.12) (0.01-3.16) (0.0-4.27) 0.56 0.38 0.57 0 3 1 1 0 (0.89-6.36) (0.0-1.23) (0.0-16.02) (0.0-52.89) 2.73 0 0 0 5 0 0 0 (0.84-1.32) (0.94-1.55) (0.70-1.37) (1.17-2.54)a 1.06 1.22 1.00 1.76 79 66 38 28
1.47 (0.98-2.11) 1.14 (0.64-1.89) 2.12 (1.16-3.56)a 29 15 14 1.10 (0.71-1.63) 0.75 (0.39-1.31) 1.97 (1.05-3.37)a 25 12 13 0.96 (0.69-1.30) 0.84 (0.55-1.23) 1.27 (0.71-2.09) 41 26 15 0.69 (0.28-1.42) 0.59 (0.16-1.50) 0.90 (0.19-2.63) 7 4 3 1.35 (0.49-2.93) 0.94 (0.19-2.74) 2.38 (0.49-6.97) 6 3 3 Overall White Nonwhite
4
$5 y F/U $2 testsb $2 tests, whites $2 tests, nonwhites
SIR (95% CI)
1.08 (0.89-1.31) 0.85 (0.65-1.10) 1.62 (1.19-2.15)a
SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases No. of ESRD Cases
108 60 48
SIR (95% CI) SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases
25-,40 mg/dL 5-,25 mg/dL 0-,5 mg/dL Overall
Highest Lead Category Achieved
Table 2. ESRD SIRs by Lead Exposure Category in the Subcohort With Data for Race
40-51 mg/dL
No. of ESRD Cases
.51 mg/dL
Chowdhury et al
more marked for nonwhites (SIR, 2.12; 95% CI, 1.16-3.56; 14 cases) than whites (SIR, 1.14; 95% CI, 0.64-1.89). Nonwhites also had a significant elevated risk in the 40- to 51-mg/dL category (SIR, 1.97; 95% CI, 1.05-3.37). When restricting the cohort to those with at least 5 years’ follow-up, the elevation in ESRD incidence in the highest category (races combined) was more marked (SIR, 1.56; 95% CI, 1.022.29; 26 cases). Table 2 also lists results for those who had 2 or more blood tests (51% of the cohort), restricted to those with 5 or more years’ follow-up, for both those with known race and for the full cohort with imputed race. This group is likely to exclude anyone with a nonoccupational blood test because men with nonoccupational blood tests are less likely to have had more than one test. Here, the elevation of risk in the highest blood lead category is even more marked (SIR, 1.74; 95% CI, 1.14-2.55), again especially among nonwhites, although whites also had an excess risk (SIR, 1.39; 95% CI, 0.76-2.33). For internal analyses of the cohort with race data and with no restriction on length of follow-up, we combined categories 1 and 2, both of which had a small number of cases, to serve as the referent. Rate ratios from Poisson regression (internal analysis) across categories 0-,25 (categories 1-2 combined), 25-,40, 40-51, and .51 mg/dL were 1.0, 1.07, 1.21, and 1.55, respectively, with a test for linear trend of P 5 0.08. Internal analyses with Cox regression yielded similar results (rate ratios of 1.00, 1.04, 1.16, and 1.54, respectively; P for trend 5 0.2). Table 3 lists results for the entire cohort using the imputed data for race. Results in Table 3 are largely consistent with results for the subcohort with data for race. The overall rate ratio was not remarkable (0.92), but again, we see an excess of ESRD incidence in the highest blood lead category (SIR, 1.36; 95% CI, 0.991.73). However, there also was an excess incidence of ESRD in the lowest blood lead category (SIR, 1.31; 95% CI, 0.82-1.79). In those with 5 or more years’ follow-up, there again is an excess incidence of ESRD in the highest blood lead category (SIR, 1.43; 95% CI, 1.01-1.85), again slightly more pronounced in nonwhites. There no longer is an excess incidence of ESRD in the lowest lead category, which has a relatively small number of ESRD cases because followup was shortest in this group (Table 1). Table 3 also shows results for those with more than one blood test, with a slightly more marked excess incidence of ESRD in the highest blood lead group (SIR, 1.59; 95% CI, 1.16-2.13). In internal analysis for the group with 5 or more years’ follow-up, we collapsed blood lead categories 1 and 2 for a test for trend because observed cases were few (n 5 9) in category 1. Using Poisson Am J Kidney Dis. 2014;-(-):---
1.59 (1.16-2.13)a
regression, rate ratios across blood lead categories (again averaging across 5 imputed data sets) were 1.0 (categories 1 and 2 combined), 0.80, 1.03, and 1.72, respectively, with a positive trend (P 5 0.003), driven by the elevation in the highest blood lead category. Internal analyses with Cox regression yielded similar results (rate ratios of 1.00, 0.92, 1.08, and 1.96; P trend 5 0.003).
Note: N 5 58,307, race imputed for 69%. Abbreviations: CI, confidence interval; ESRD, end-stage renal disease; F/U, follow-up; SIR, standardized incidence ratio. a Statistically significant at P # 0.05. b There was 51% of the cohort with more than 1 blood test. Men with more than 1 blood test are likely to exclude any nonoccupational cases.
1.43 (1.01-1.85)a
44 134 $5 y F/U, $2 testsb
0.95 (0.79-1.12)
1
0.93 (0.02-5.21)
12
0.77 (0.39-1.34)
46
0.75 (0.55-0.99)
31
0.87 (0.59-1.23)
1.36 (0.99-1.73)
45 0.85 (0.58-1.12)
52 44
38 0.64 (0.49-0.80)
0.73 (0.59-0.88) 98
65 0.73 (0.49-0.98)
1.00 (0.78-1.23) 79
34 1.03 (0.36-1.71)
1.31 (0.83-1.79) 29
9 0.82 (0.71-0.94) 191 $5 y F/U
SIR (95% CI)
0.92 (0.82-1.03) 302 Overall
SIR (95% CI) SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases No. of ESRD Cases
Overall
Am J Kidney Dis. 2014;-(-):---
0.80 (0.56-1.03)
SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases No. of ESRD Cases
25-,40 mg/dL 5-,25 mg/dL 0-,5 mg/dL
Highest Lead Category Achieved
Table 3. ESRD SIRs by Blood Lead Category in Full Cohort
40-51 mg/dL
.51 mg/dL
Lead Exposure and Incident ESRD
DISCUSSION We found evidence of increased ESRD incidence in this lead-exposed cohort, restricted primarily to the highest blood lead category (.51 mg/dL). We found this evidence in both analyses restricted to the 31% of the cohort with known race and the entire cohort for which we imputed missing race. Compared to the US population, the risk of ESRD increased 40%-50% in the highest blood lead category in those with 5 or more years’ follow-up. The more pronounced excess incidence of ESRD in those with more follow-up is consistent with the idea that chronic diseases typically require some latency period, such that effects are delayed and long term. However, given our lack of detailed work history, we do not know the true date of initial lead exposure. We also found some excess risk in the lowest blood lead category (0-,5 mg/dL). However, in the portion of the cohort in which race data were available, the numbers of ESRD cases were small, and CIs for rate ratios are wide (Table 2). In Table 3 (full cohort, race imputed), numbers were larger, but the elevation was no longer present in the analysis with 5 or more years’ follow-up. It is possible that the elevated rate ratio here was due to reverse causation, whereby someone with decreased kidney function just before ESRD diagnosis had their blood lead level measured and found to be high because the kidney was not excreting lead properly. The same phenomenon might have occurred for the 61 cases excluded because their kidneys failed before their first blood lead test, in other words, the high blood lead level was found pursuant to testing for poor kidney function. However, the evidence for reverse causation is mixed.36 In most analyses, we found no excess risk in categories of lead below the highest category of .51 mg/dL, which is surprising in that Ekong et al10 recently found effects on glomerular function at much lower levels, and other experts have called for lowering OSHA’s blood lead standard to levels much lower than the current limits (to ,20 mg/dL).37-39 One possible explanation of not finding effects at lower levels is that we had no data for duration of exposure. If cumulative exposure is the underlying metric of interest in gauging risk, often the case for chronic disease, we may have misclassified individuals, using results of one or a few blood lead tests to characterize 5
Chowdhury et al
their exposure. This may have caused us to miss excess risk in subsets of those with lower blood lead levels who had longer durations of exposure. It also is worth noting in this context that we had no data for kidney damage except for ESRD, which otherwise might have permitted us to see effects at lower exposures. It is possible that finding an excess of ESRD incidence in the highest lead category reflects the wearing off of the healthy worker effect, given that follow-up was greatest for those in the highest blood lead categories.40 However, there was little evidence of an overall healthy worker effect for the entire cohort (SIRs of 0.92 in the whole cohort and 1.08 in the subcohort with known race). We had to rely on a few blood lead test results to categorize exposure. However, we believe it is likely that those with high blood lead test results usually will have had higher overall average exposure than those with lower blood lead test results; that is, the relative ranking of individuals by blood lead category is likely to be reasonably accurate. The excess risk in the highest category was more pronounced for nonwhites than whites. Our finding may imply that high lead exposure exacerbates the already high underlying risk for the nonwhite group. Our findings for excess risk in the group with highest blood lead levels, although limited, are in accord with the literature that high levels are associated with chronic kidney disease in both animals and humans.10 Much of the evidence for humans supporting an association between high exposure and chronic kidney disease comes from occupational mortality studies with few deaths.12-15 Not all studies of humans have been positive. A recent case-control study by Evans et al41 found no association with estimated levels of lead exposure; however, few individuals were occupationally exposed. Our findings are consistent with the earlier findings of cohort studies in which exposures were known to be high and considerable time has passed since first exposure; our data are based on larger numbers and based on incidence rather than mortality. The highest blood lead category used in our study, which constituted 7% of the cohort and included those with blood lead levels . 51 mg/dL, is beyond the 50-mg/dL threshold at which OSHA requires removal of a construction worker to a lower exposed job until his or her blood lead level is lowered to ,40 mg/dL (for other workers, these limits are $60 mg/dL on a single test or a series of recent tests averaging .50 mg/dL).6 In our more recent data (2000-2009; n 5 49,000 tests), blood test results of 40-,51 and .51 mg/dL represented 4.3% and 2.1% of tests, respectively. Given that repeated intermittent high exposures can lead to a high body burden over time, 6
which may lead to kidney damage, the precautionary principle would suggest that the present OSHA standard, even if consistently enforced, may not be sufficiently protective. Weaknesses in our data include lack of a complete work history, lack of data for possible confounders (eg, hypertension and diabetes), missing data for race for most of the cohort, and possibly inaccurate matching with USRDS and NDI registries due to the lack of social security number information in most of our cohort. However, previous work by Williams et al42 has shown that with name and date of birth, investigators can attain 92% sensitivity and 92% specificity using NDI, a result also likely to be true for USRDS. We have studied a large cohort with occupational exposure to lead; lead is a relatively common exposure among adult workers. We found evidence of increased ESRD incidence at high blood levels, primarily among nonwhites. Data are limited by the lack of detailed work history and reliance on only a few blood lead tests to estimate level of exposure. Strengths of our study include a large cohort, documented blood lead levels, and use of ESRD incidence, rather than reliance on chronic kidney disease mortality. This is a young cohort, and further follow-up is needed to confirm our findings.
ACKNOWLEDGEMENTS We acknowledge the following people for help providing data access and matching: Susan Payne (California Department of Health Services), Beth Forrest (USRDS), Michele Goodier (NDI); ABLES Program: Walter Alarcon (NIOSH), Kenneth D. Rosenman (Michigan State University), Thomas St. Louis (Connecticut Department of Health), Henry A. Anderson, MD (Wisconsin Department of Health and Family Services), Jay Devasundaram (Pennsylvania Department of Health), Ed Socie, PhD (Ohio Department of Health), David J. Valiante (New Jersey Department of Health and Senior Services), Alicia Fletcher (New York Department of Health), Rita Gergely (Iowa Department of Health), Erik Zabel (Minnesota Department Health), and Bob Nicotera (Massachusetts Department of Labor). Support: This work was supported by NIOSH award R01OH00898. Financial Disclosure: The authors declare that they have no other relevant financial interests.
SUPPLEMENTARY MATERIAL Item S1: Race imputation methods. Note: The supplementary material accompanying this article (http://dx.doi.org/10.1053/j.ajkd.2013.12.005) is available at www.ajkd.org
REFERENCES 1. US Environmental Protection Agency. Air Quality Criteria for Lead (Final). EPA-600/8-77/017. Research Triangle Park, NC: National Center for Environmental Assessment-RTP Office, Office of Research and Development, US Environmental Protection Agency; 1977. Am J Kidney Dis. 2014;-(-):---
Lead Exposure and Incident ESRD 2. Pirkle JL, Brody DJ, Gunter EW, et al. The decline in blood lead levels in the United States. The National Health and Nutrition Examination Surveys (NHANES). JAMA. 1994;272(4):284-291. 3. US Environmental Protection Agency. Outdoor air. Exhibit 2.5. Lead emissions in the US by source category 1970-2005. http://www. epa.gov/ncea/roe/index.htm. Accessed January 6, 2014. 4. Staudinger KC, Roth VS. Occupational lead poisoning. Am Fam Physician. 1998;57(4):719-726. 731-712. 5. Centers for Disease Control and Prevention. Blood lead level data, 2002-2005. http://www.cdc.gov/niosh/topics/ABLES/pdfs/2 002-2005lead_data.pdf. Accessed January 6, 2014. 6. Occupational Safety and Health Administration. Lead standards. https://www.osha.gov/SLTC/lead. Accessed January 6, 2014. 7. Levey AS, Stevens LA, Schmid CH, et al. CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604-612. 8. Collins AJ, Foley RN, Herzog C, et al. US Renal Data System 2012 annual data report. Am J Kidney Dis. 2013;61(1)(suppl 1):e1-e480. 9. US Environmental Protection Agency. Air Quality Criteria for Lead (2006) Final Report. EPA/600/R-05/144aF-bF. Research Triangle Park, NC: National Center for Environmental AssessmentRTP Office, Office of Research and Development; 2006. 10. Ekong E, Jaar B, Weaver V. Lead-related nephrotoxicity: a review of the epidemiologic evidence. Kidney Int. 2006;70(12): 2074-2084. 11. Evans M, Elinder CG. Chronic renal failure from lead: myth or evidence-based fact? Kidney Int. 2011;79(3):272-279. 12. Cocco P, Hua F, Boffetta P, et al. Mortality of Italian lead smelter workers. Scand J Work Environ Health. 1997;23(1):15-23. 13. Steenland NK, Thun MJ, Ferguson CW, Port FK. Occupational and other exposures associated with male end-stage renal disease: a case/control study. Am J Public Health. 1990;80(2):153-157. 14. Cooper W, Wong O, Kheifets L. Mortality among employees of lead battery plants and lead-producing plants, 19471980. Scand J Work Environ Health. 1985;11(5):331-345. 15. Steenland K, Selevan S, Landrigan P. The mortality of lead smelter workers: an update. Am J Public Health. 1992;82(12): 1641-1644. 16. Staessen JA, Lauwerys RR, Buchet JP, et al. Impairment of renal function with increasing blood lead concentrations in the general population. The Cadmibel Study Group. N Engl J Med. 1992;327(3):151-156. 17. Payton M, Hu H, Sparrow D, Weiss ST. Low-level lead exposure and renal function in the Normative Aging Study. Am J Epidemiol. 1994;140(9):821-829. 18. Chia KS, Jeyaratnam J, Tan C, Ong HY, Ong CN, Lee E. Glomerular function of lead-exposed workers. Toxicol Lett. 1995;77(1-3):319-328. 19. Kim R, Rotnitsky A, Sparrow D, Weiss S, Wager C, Hu H. A longitudinal study of low-level lead exposure and impairment of renal function. The Normative Aging Study. JAMA. 1996;275(15): 1177-1181. 20. Lin JL, Tan DT, Hsu KH, Yu CC. Environmental lead exposure and progressive renal insufficiency. Arch Intern Med. 2001;161(2):264-271. 21. Lin JL, Yu CC, Lin-Tan DT, Ho HH. Lead chelation therapy and urate excretion in patients with chronic renal diseases and gout. Kidney Int. 2001;60(1):266-271. 22. Weaver VM, Lee BK, Ahn KD, et al. Associations of lead biomarkers with renal function in Korean lead workers. Occup Environ Med. 2003;60(8):551-562.
Am J Kidney Dis. 2014;-(-):---
23. Tsaih SW, Korrick S, Schwartz J, et al. Lead, diabetes, hypertension, and renal function: the Normative Aging Study. Environ Health Perspect. 2004;112(11):1178-1182. 24. Lin JL, Lin-Tan DT, Li YJ, Chen KH, Huang YL. Lowlevel environmental exposure to lead and progressive chronic kidney diseases. Am J Med. 2006;119(8):707 e701-709. 25. Weaver VM, Griswold M, Todd AC, et al. Longitudinal associations between lead dose and renal function in lead workers. Environ Res. 2009;109(1):101-107. 26. Navas-Acien A, Tellez-Plaza M, Guallar E, et al. Blood cadmium and lead and chronic kidney disease in US adults: a joint analysis. Am J Epidemiol. 2009;170(9):1156-1164. 27. Roscoe RJ, Ball W, Curran JJ, et al. Adult blood lead epidemiology and surveillance—United States, 1998-2001. MMWR Surveill Summ. 2002;51(11):1-10. 28. Centers for Disease Control and Prevention. Adult blood lead epidemiology and surveillance—United States, 2008-2009. MMWR Morbid Mortal Wk Rep. 2011;60(25):841-845. 29. USRDS. Completeness and reliability of USRDS data: comparisons with the Michigan Kidney Registry. Am J Kidney Dis. 1992;20(5)(suppl 2):84-88. 30. Calvert GM, Steenland K, Palu S. End-stage renal disease among silica-exposed gold miners. A new method for assessing incidence among epidemiologic cohorts. JAMA. 1997;277(15): 1219-1223. 31. Steenland K, Sanderson W, Calvert GM. Kidney disease and arthritis in a cohort study of workers exposed to silica. Epidemiology. 2001;12(4):405-412. 32. Radican L, Wartenberg D, Rhoads GG, et al. A retrospective occupational cohort study of end-stage renal disease in aircraft workers exposed to trichloroethylene and other hydrocarbons. J Occup Environ Med. 2006;48(1):1-12. 33. US Renal Data System. ADR reference tables. http://www. usrds.org/reference.aspx. Accessed June 9, 2013. 34. Robinson CF, Schnorr TM, Cassinelli RT II, et al. Tenth revision U.S. mortality rates for use with the NIOSH Life Table Analysis System. J Occup Environ Med. 2006;48(7):662-667. 35. Schubauer-Berigan MK, Hein MJ, Raudabaugh WM, et al. Update of the NIOSH life table analysis system: a person-years analysis program for the windows computing environment. Am J Ind Med. 2011;54(12):915-924. 36. Fadrowski JJ, Abraham AG, Navas-Acien A, Guallar E, Weaver VM, Furth SL. Blood lead level and measured glomerular filtration rate in children with chronic kidney disease. Environ Health Perspect. 2013;121(8):965-970. 37. Kosnett MJ, Wedeen RP, Rothenberg SJ, et al. Recommendations for medical management of adult lead exposure. Environ Health Perspect. 2007;115(3):463-471. 38. Schwartz BS, Hu H. Adult lead exposure: time for change. Environ Health Perspect. 2007;115(3):451-454. 39. Spivey A. The weight of lead. Effects add up in adults. Environ Health Perspect. 2007;115(1):A30-A36. 40. Checkoway HPN, Kriebel D. Research Methods in Occupational Epidemiology. New York, NY: Oxford University Press; 2004. 41. Evans M, Fored CM, Nise G, Bellocco R, Nyrén O, Elinder CG. Occupational lead exposure and severe CKD: a population-based case-control and prospective observational cohort study in Sweden. Am J Kidney Dis. 2010;55(3):497-506. 42. Williams BC, Demitrack LB, Fries BE. The accuracy of the National Death Index when personal identifiers other than Social Security number are used. Am J Public Health. 1992;82(8): 1145-1147.
7
L
ead exposure in the United States has declined since the 1970s.1 Blood lead levels in the United States decreased from 12.2 to 2.8 mg/dL from 1976 to 19912 and then to 1.3 mg/dL in 2008.3 However, some US adults still have high occupational exposure to lead. The National Institute for Occupational Safety and Health (NIOSH) estimated in the 1980s that more than 3 million workers in the United States were exposed.4 NIOSH currently conducts an adult blood lead surveillance program that captures those with higher exposure; data from 37 states indicated that approximately 130,000 adults had been tested for blood lead in 2005.5 The current US Occupational Safety and Health Administration (OSHA) standard calls for workers to be removed from exposure when they have: (1) a blood lead level of 50 mg/dL (construction workers) or (2) 60 mg/dL on a single test or an average of $50 mg/dL on a series of recent tests (other workers) and to not return until their blood lead level decreases to ,40 mg/dL.6 Twelve percent of US adults have diagnosed chronic kidney disease, with increasing numbers each year; 6% have stage 3 disease or higher Am J Kidney Dis. 2014;-(-):---
(glomerular filtration rate , 60 mL/min/1.73 m2).7 There currently are about 600,000 prevalent cases of end-stage renal disease (ESRD) in the United States.8 Very high lead exposure historically has been shown to lead to an increased risk of acute kidney injury.9 Chronic high levels of exposure resulting in blood lead levels . 70-80 mg/dL have been associated with lead nephropathy, characterized by tubular damage, in studies10 of both humans and animals, although some researchers are unconvinced by this evidence.11 Much of the evidence for humans for chronic kidney disease From the Departments of 1Epidemiology and 2Environmental Health, Rollins School of Public Health, Emory University, Atlanta, GA. Received September 1, 2013. Accepted in revised form December 4, 2013. Address correspondence to Kyle Steenland, PhD, Department of Epidemiology, Rollins School of Public Health, Emory University, 1518 Clifton Rd, Atlanta, GA 30322. E-mail: nsteenl@ emory.edu 2014 by the National Kidney Foundation, Inc. 0272-6386/$36.00 http://dx.doi.org/10.1053/j.ajkd.2013.12.005 1
Chowdhury et al
in relation to high lead exposures comes from occupational studies.12-15 In recent years, there also has been increasing evidence of lead’s negative effect on kidney function (eg, increased serum creatinine levels) at lower levels of lead exposure than studies in earlier occupational cohorts.10,16-26 Here, we have studied the incidence of ESRD in men tested for blood lead in 11 US states as part of the NIOSH-sponsored Adult Blood Lead Epidemiology and Surveillance (ABLES) program.
METHODS Data Sources and Study Participants The ABLES program started collecting state-level data for blood lead exposure in 1987.27 Initially, some states gathered data only for individuals with blood lead levels . 25 mg/dL, but subsequently began to collect data at lower levels. Blood lead tests were performed primarily because of occupational exposure. ABLES coverage increased from 4 states in 1987 to 41 states in 2012.27 NIOSH has collected data regarding industry for a limited number of ABLES individuals (n 5 6,999) with blood lead levels $ 25 mg/dL.28 Of these, 62% were in manufacturing (43% in battery production or renovation), 10% were in construction, 7% were in metal mining, and 21% were in other industries or data were unavailable. Only 4% of these individuals were exposed nonoccupationally (eg, by lead in shooting ranges). We obtained data from 11 state ABLES programs— Connecticut, California, Ohio, Minnesota, Iowa, Pennsylvania, New York, New Jersey, Wisconsin, Michigan, and Massachusetts— from their year of first participation through 2008. These states made up 4%, 18%, 10%, 2%, 3%, 3%, 29%, 11%, 6%, 3%, and 11%, respectively, of the overall cohort. These states were chosen because they had the most individuals with blood lead data and had data that went back the farthest in time. Furthermore, adding additional states would not have been possible due to the increased costs and our budget constraints. We excluded everyone first tested after 2005 to avoid short follow-up. We excluded those missing information for date of birth, test date, or blood lead levels. We categorized each blood lead level reading into 1 of 5 categories: 0-,5, 5-,25, 25-,40, 40-51, and .51 mg/dL. Categories of ,25, 25-40, and .40 mg/dL traditionally have been used to categorize occupational blood lead levels, whereas the lowest category we used (,5 mg/dL) essentially is equivalent to nonoccupational US blood lead levels. We divided the upper category (.40 mg/dL) into 2 groups with an approximately equal number of ESRD cases to explore whether there was excess risk confined to those with particularly high blood lead levels. For individuals with more than 1 blood lead test, blood lead category was defined as the highest category ever achieved. Among those eligible for inclusion, we first selected everyone who had ever had a blood lead level $ 25 mg/dL (categories 3-5). We then randomly selected an equal number of people from categories 1 and 2 (50% from each category), stratified by state. Three states (Massachusetts, Michigan, and Wisconsin) opted to do their own data processing and matching with the National Death Index (NDI) and the US Renal Data System (USRDS). They followed the same selection pattern and sent us de-identified data. We restricted our analytic cohort to men because women represented only 15% of the cohort and only 9% of ESRD cases and were concentrated more heavily in the lowest blood lead category (40% had levels , 5 mg/dL compared to 11% for men), 2
suggesting that they may have included a higher percentage of nonoccupational blood leads, for example, pregnant women tested for lead during pregnancy (Susan Payne, California ABLES/ Occupational Lead Poisoning Prevention Program, personal communication, May 2013). We further excluded all people who were tested for the first time after the age of 70 years or before the age of 18 years and any implausible blood test results (.250 mg/dL). We excluded some ESRD cases (n 5 61) because their outcomes had occurred (ESRD diagnosis) before or on the same day as their first blood lead measurement. After the mentioned exclusions, we had a final analytic data set with 58,307 individuals.
Matching With NDI (deaths) and USRDS (incident ESRD) Matching with the NDI was done to determine vital status. We used first name, last name, date of birth, sex, race (when available), and social security number (when available) to match with the NDI through 2010 (2008 for Massachusetts and 2009 for Michigan and Wisconsin, all of which did their own matching). To determine a true match among possible multiple NDI matches, we selected those assigned a status code of 1 by NDI, indicating a high probability of a match. Past validation studies indicate that 95% of patients with ESRD are in the USRDS29; good records are kept, given the high cost of treatment, which is covered by Medicare or the US Department of Veterans Affairs. We used the same matching variables as we did for NDI for USRDS matching through 2010 (2008 for Michigan and Wisconsin). We accepted all matches so designated by the USRDS. Similar matching of other occupational cohorts has been done in the past.30-32
Adjustment for Missing Race The majority of our cohort was missing data for race (69%). Nonwhites in the United States have much higher ESRD rates than whites. Age- and sex-adjusted ESRD incidence rates (per 100,000) for African Americans and whites in 2010 in the United States were 92.4 and 27.5, respectively.33 Although race was not associated with lead level among those with known race (Table 1) and would not be expected to affect internal analyses, misclassified race could result in incorrect expected ESRD cases in external analyses (observed cases are unaffected because we had data for race for all of them). We chose 2 strategies to deal with the problem of missing race. First, we conducted an analysis restricted to the 31% with known race. Second, we imputed race for those missing race (although not for the 302 ESRD cases in the total cohort, for whom race was known) by multiple imputation (n 5 5) and ran a life table for each imputation. Details can be found in Item S1 (provided as online supplementary material). Briefly, we built a model that predicted race among those with data for race and used that model to predict race for those without data for race. The model correctly predicted race for w70% of individuals; given that race is known for 31% of the cohort, we can expect that race will be misclassified for w20% of the full cohort using imputation. Given that whites will be misclassified more than nonwhites, this remaining misclassification likely would cause some overestimation of expected ESRD cases in the full cohort analysis, resulting in conservative underestimates of the true rate ratios for exposed versus nonexposed (see Item S1 for details).
Analyses The NIOSH Life Table Analysis System was used to calculate person-years at risk and rates of ESRD incidence for the cohort and then compare these rates with those of the US population by standardized incidence ratios (SIRs), adjusted for age, race, sex, and calendar year.34,35 Am J Kidney Dis. 2014;-(-):---
Lead Exposure and Incident ESRD Table 1. Description of Demographics by Lead Category Highest Lead Category Achieved Total
0-,5 mg/dL
5-,25 mg/dL
25-,40 mg/dL
40-51 mg/dL
.51 mg/dL
58,307
6,832 (12%)
18,618 (32%)
21,440 (37%)
7,348 (12%)
4,069 (7%)
12.2
6.4
10.2
14.3
17
17.7
39.0 6 11.7
40.7 6 12.2
40.0 6 11.7
37.9 6 11.4
38.0 6 11.4
38.9 6 11.7
15,936 (27.3%) 2,315 (4.0%) 40,056 (68.7%) 0.13%
1,499 (22%) 222 (3%) 5,111 (75%) 0.13%
2,566 (14%) 416 (2%) 15,636 (84%) 0.14%
7,015 (32.7%) 954 (4.4%) 13,471 (62.8%) 0.12%
2,981 (4.06%) 423 (5.8%) 3,944 (53.7%) 0.12%
1,875 (46.1%) 300 (7.4%) 1,894 (46.6%) 0.14%
No. with single tests
28,540 (48.9%)
6,108 (89.4%)
12,710 (68.3%)
7,783 (36.3%)
1,445 (19.7%)
494 (12.1%)
Median no. of testsb
4
2
3
4
6
7
25.9 6 17.9
2.5 6 1.1
13.0 6 5.5
30.8 6 4.2
44.7 6 3.3
65.3 6 18.7
No. of participants Median follow-up (y) Mean age at first testa (y) Race White Nonwhite Missing/unknown % nonwhite among those with known race
Mean highest blood lead level (mg/dL)a No. with SSN data
15,233 (26.1%)
611 (8.9%)
2,079 (11.2%)
7,661 (35.7%)
3,104 (42.2%)
1,778 (43.7%)
Median year of birth
1959
1962
1961
1959
1956
1954
Median year of ESRD
2004
2007
2004
2004
2003
2003
302 (0.5%)
29 (0.4%)
79 (0.4%)
98 (0.5%)
44 (0.6%)
52 (1.2%)
No. with ESRD
Abbreviations: ESRD, end-stage renal disease; SSN, social security number. a Expressed as mean 6 standard deviation. b In those with more than 1 test.
Incidence rates of ESRD in the United States are available since 1973 and presently go through 2007. We used these US ESRD rates, extrapolated until 2010 (the end of our follow-up), as the referent rates to calculated expected ESRD cases, adjusted for age, race, sex, and calendar year. We calculated SIRs for all blood lead categories (1-5) and all categories combined, as well as by time since first exposure (0-5 or $5 years). Person-time at risk began at the first blood test. All person-years were assigned to the highest blood level ever attained regardless of the number of tests. For example, if a man had 3 tests with blood levels of 35, 43, and 48 mg/dL, he was assigned to the category 40-51 mg/dL, and all his person-time (beginning at his first blood test) was assigned to that category. Most people (68%) did not change blood lead category, whereas 24% changed category by only one category. Only 8% of the cohort changed blood lead category by more than one blood lead level. Person-time ended for everyone in 2010, the end of our NDI and USRDS follow-up, except for Massachusetts (2009), Michigan (2008), and Wisconsin (2008). In internal analyses, we calculated rate ratios using Poisson models (log person-years offset, correction for overdispersion) adjusted for sex, race, age, and calendar period (SAS, version 9.3; SAS Institute Inc). We compared lead categories 3-5 (25-,40, 40-51, and .51 mg/dL) to a reference category (,25 mg/dL). We combined categories 1 and 2 (0-,5 and 5-,25 mg/dL) due to the small number of cases in each of them. These analyses were based on the person-years calculated by the life table analyses and took into account the censoring occurring with mortality. We also conducted a trend test for ESRD incidence by assigning median values for each blood lead category in the regression, such that lead level became a continuous rather than a categorical variable. The test for trend was based on the significance of that variable. Finally, we also conducted internal analyses using Cox regression. In these analyses, we constructed 100-person risk sets for each case, randomly selected from those who attained the same age or older as their case’s age at diagnosis and restricted to those Am J Kidney Dis. 2014;-(-):---
who were born 6 2.5 years from the date of birth of their respective cases.
RESULTS Table 1 provides descriptive information about the cohort. There were 58,307 men. Forty-nine percent had only one blood test, whereas the rest had a median of 4 tests. Median follow-up was 12 years (increasing from 6.4 years in the lowest blood lead category to 17.7 years in the highest). There were 3,337 deaths in the cohort. The percentage of nonwhites was similar across blood lead categories, among those with known race. We did not have data regarding occupation or industry for the entire cohort. However, we had data for 72% of our California cohort (w7,600 men) for whether a blood test was occupational or nonoccupational (eg, from exposure to lead at a shooting range or lead paint in a residence). Of these, only 2% were nonoccupational. In the highest blood lead category, 82% had information on occupation, compared to 48% in the lowest category. Table 2 shows a comparison of incident ESRD in the cohort by lead category using the US population as referent (external comparison). Overall, there was a slight excess of ESRD incidence (SIR, 1.08; 95% confidence interval [CI], 0.89-1.31; 108 cases) compared to the general population. There was some elevation in ESRD incidence in the highest blood lead category (SIR, 1.47; 95% CI, 0.98-2.11), which was 3
(1.02-2.29)a (1.14-2.55)a (0.76-2.33) (1.28-4.32)a 1.56 1.74 1.39 2.48 26 26 14 12 Note: n 5 18,251. Abbreviations: CI, confidence interval; ESRD, end-stage renal disease; F/U, follow-up; SIR, standardized incidence ratio. a Statistically significant at P # 0.05. b There was 51% of the cohort with more than 1 blood test. Men with more than 1 blood test are likely to exclude any nonoccupational cases.
(0.70-1.73) (0.72-1.93) (0.40-1.65) (0.94-3.91) 1.13 1.22 0.87 2.06 21 18 9 9 (0.48-1.11) (0.60-1.49) (0.49-1.49) (0.49-2.49) 0.74 0.98 0.89 1.21 24 21 14 7 (0.11-1.62) (0.01-2.12) (0.01-3.16) (0.0-4.27) 0.56 0.38 0.57 0 3 1 1 0 (0.89-6.36) (0.0-1.23) (0.0-16.02) (0.0-52.89) 2.73 0 0 0 5 0 0 0 (0.84-1.32) (0.94-1.55) (0.70-1.37) (1.17-2.54)a 1.06 1.22 1.00 1.76 79 66 38 28
1.47 (0.98-2.11) 1.14 (0.64-1.89) 2.12 (1.16-3.56)a 29 15 14 1.10 (0.71-1.63) 0.75 (0.39-1.31) 1.97 (1.05-3.37)a 25 12 13 0.96 (0.69-1.30) 0.84 (0.55-1.23) 1.27 (0.71-2.09) 41 26 15 0.69 (0.28-1.42) 0.59 (0.16-1.50) 0.90 (0.19-2.63) 7 4 3 1.35 (0.49-2.93) 0.94 (0.19-2.74) 2.38 (0.49-6.97) 6 3 3 Overall White Nonwhite
4
$5 y F/U $2 testsb $2 tests, whites $2 tests, nonwhites
SIR (95% CI)
1.08 (0.89-1.31) 0.85 (0.65-1.10) 1.62 (1.19-2.15)a
SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases No. of ESRD Cases
108 60 48
SIR (95% CI) SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases
25-,40 mg/dL 5-,25 mg/dL 0-,5 mg/dL Overall
Highest Lead Category Achieved
Table 2. ESRD SIRs by Lead Exposure Category in the Subcohort With Data for Race
40-51 mg/dL
No. of ESRD Cases
.51 mg/dL
Chowdhury et al
more marked for nonwhites (SIR, 2.12; 95% CI, 1.16-3.56; 14 cases) than whites (SIR, 1.14; 95% CI, 0.64-1.89). Nonwhites also had a significant elevated risk in the 40- to 51-mg/dL category (SIR, 1.97; 95% CI, 1.05-3.37). When restricting the cohort to those with at least 5 years’ follow-up, the elevation in ESRD incidence in the highest category (races combined) was more marked (SIR, 1.56; 95% CI, 1.022.29; 26 cases). Table 2 also lists results for those who had 2 or more blood tests (51% of the cohort), restricted to those with 5 or more years’ follow-up, for both those with known race and for the full cohort with imputed race. This group is likely to exclude anyone with a nonoccupational blood test because men with nonoccupational blood tests are less likely to have had more than one test. Here, the elevation of risk in the highest blood lead category is even more marked (SIR, 1.74; 95% CI, 1.14-2.55), again especially among nonwhites, although whites also had an excess risk (SIR, 1.39; 95% CI, 0.76-2.33). For internal analyses of the cohort with race data and with no restriction on length of follow-up, we combined categories 1 and 2, both of which had a small number of cases, to serve as the referent. Rate ratios from Poisson regression (internal analysis) across categories 0-,25 (categories 1-2 combined), 25-,40, 40-51, and .51 mg/dL were 1.0, 1.07, 1.21, and 1.55, respectively, with a test for linear trend of P 5 0.08. Internal analyses with Cox regression yielded similar results (rate ratios of 1.00, 1.04, 1.16, and 1.54, respectively; P for trend 5 0.2). Table 3 lists results for the entire cohort using the imputed data for race. Results in Table 3 are largely consistent with results for the subcohort with data for race. The overall rate ratio was not remarkable (0.92), but again, we see an excess of ESRD incidence in the highest blood lead category (SIR, 1.36; 95% CI, 0.991.73). However, there also was an excess incidence of ESRD in the lowest blood lead category (SIR, 1.31; 95% CI, 0.82-1.79). In those with 5 or more years’ follow-up, there again is an excess incidence of ESRD in the highest blood lead category (SIR, 1.43; 95% CI, 1.01-1.85), again slightly more pronounced in nonwhites. There no longer is an excess incidence of ESRD in the lowest lead category, which has a relatively small number of ESRD cases because followup was shortest in this group (Table 1). Table 3 also shows results for those with more than one blood test, with a slightly more marked excess incidence of ESRD in the highest blood lead group (SIR, 1.59; 95% CI, 1.16-2.13). In internal analysis for the group with 5 or more years’ follow-up, we collapsed blood lead categories 1 and 2 for a test for trend because observed cases were few (n 5 9) in category 1. Using Poisson Am J Kidney Dis. 2014;-(-):---
1.59 (1.16-2.13)a
regression, rate ratios across blood lead categories (again averaging across 5 imputed data sets) were 1.0 (categories 1 and 2 combined), 0.80, 1.03, and 1.72, respectively, with a positive trend (P 5 0.003), driven by the elevation in the highest blood lead category. Internal analyses with Cox regression yielded similar results (rate ratios of 1.00, 0.92, 1.08, and 1.96; P trend 5 0.003).
Note: N 5 58,307, race imputed for 69%. Abbreviations: CI, confidence interval; ESRD, end-stage renal disease; F/U, follow-up; SIR, standardized incidence ratio. a Statistically significant at P # 0.05. b There was 51% of the cohort with more than 1 blood test. Men with more than 1 blood test are likely to exclude any nonoccupational cases.
1.43 (1.01-1.85)a
44 134 $5 y F/U, $2 testsb
0.95 (0.79-1.12)
1
0.93 (0.02-5.21)
12
0.77 (0.39-1.34)
46
0.75 (0.55-0.99)
31
0.87 (0.59-1.23)
1.36 (0.99-1.73)
45 0.85 (0.58-1.12)
52 44
38 0.64 (0.49-0.80)
0.73 (0.59-0.88) 98
65 0.73 (0.49-0.98)
1.00 (0.78-1.23) 79
34 1.03 (0.36-1.71)
1.31 (0.83-1.79) 29
9 0.82 (0.71-0.94) 191 $5 y F/U
SIR (95% CI)
0.92 (0.82-1.03) 302 Overall
SIR (95% CI) SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases No. of ESRD Cases
Overall
Am J Kidney Dis. 2014;-(-):---
0.80 (0.56-1.03)
SIR (95% CI) SIR (95% CI)
No. of ESRD Cases No. of ESRD Cases No. of ESRD Cases
25-,40 mg/dL 5-,25 mg/dL 0-,5 mg/dL
Highest Lead Category Achieved
Table 3. ESRD SIRs by Blood Lead Category in Full Cohort
40-51 mg/dL
.51 mg/dL
Lead Exposure and Incident ESRD
DISCUSSION We found evidence of increased ESRD incidence in this lead-exposed cohort, restricted primarily to the highest blood lead category (.51 mg/dL). We found this evidence in both analyses restricted to the 31% of the cohort with known race and the entire cohort for which we imputed missing race. Compared to the US population, the risk of ESRD increased 40%-50% in the highest blood lead category in those with 5 or more years’ follow-up. The more pronounced excess incidence of ESRD in those with more follow-up is consistent with the idea that chronic diseases typically require some latency period, such that effects are delayed and long term. However, given our lack of detailed work history, we do not know the true date of initial lead exposure. We also found some excess risk in the lowest blood lead category (0-,5 mg/dL). However, in the portion of the cohort in which race data were available, the numbers of ESRD cases were small, and CIs for rate ratios are wide (Table 2). In Table 3 (full cohort, race imputed), numbers were larger, but the elevation was no longer present in the analysis with 5 or more years’ follow-up. It is possible that the elevated rate ratio here was due to reverse causation, whereby someone with decreased kidney function just before ESRD diagnosis had their blood lead level measured and found to be high because the kidney was not excreting lead properly. The same phenomenon might have occurred for the 61 cases excluded because their kidneys failed before their first blood lead test, in other words, the high blood lead level was found pursuant to testing for poor kidney function. However, the evidence for reverse causation is mixed.36 In most analyses, we found no excess risk in categories of lead below the highest category of .51 mg/dL, which is surprising in that Ekong et al10 recently found effects on glomerular function at much lower levels, and other experts have called for lowering OSHA’s blood lead standard to levels much lower than the current limits (to ,20 mg/dL).37-39 One possible explanation of not finding effects at lower levels is that we had no data for duration of exposure. If cumulative exposure is the underlying metric of interest in gauging risk, often the case for chronic disease, we may have misclassified individuals, using results of one or a few blood lead tests to characterize 5
Chowdhury et al
their exposure. This may have caused us to miss excess risk in subsets of those with lower blood lead levels who had longer durations of exposure. It also is worth noting in this context that we had no data for kidney damage except for ESRD, which otherwise might have permitted us to see effects at lower exposures. It is possible that finding an excess of ESRD incidence in the highest lead category reflects the wearing off of the healthy worker effect, given that follow-up was greatest for those in the highest blood lead categories.40 However, there was little evidence of an overall healthy worker effect for the entire cohort (SIRs of 0.92 in the whole cohort and 1.08 in the subcohort with known race). We had to rely on a few blood lead test results to categorize exposure. However, we believe it is likely that those with high blood lead test results usually will have had higher overall average exposure than those with lower blood lead test results; that is, the relative ranking of individuals by blood lead category is likely to be reasonably accurate. The excess risk in the highest category was more pronounced for nonwhites than whites. Our finding may imply that high lead exposure exacerbates the already high underlying risk for the nonwhite group. Our findings for excess risk in the group with highest blood lead levels, although limited, are in accord with the literature that high levels are associated with chronic kidney disease in both animals and humans.10 Much of the evidence for humans supporting an association between high exposure and chronic kidney disease comes from occupational mortality studies with few deaths.12-15 Not all studies of humans have been positive. A recent case-control study by Evans et al41 found no association with estimated levels of lead exposure; however, few individuals were occupationally exposed. Our findings are consistent with the earlier findings of cohort studies in which exposures were known to be high and considerable time has passed since first exposure; our data are based on larger numbers and based on incidence rather than mortality. The highest blood lead category used in our study, which constituted 7% of the cohort and included those with blood lead levels . 51 mg/dL, is beyond the 50-mg/dL threshold at which OSHA requires removal of a construction worker to a lower exposed job until his or her blood lead level is lowered to ,40 mg/dL (for other workers, these limits are $60 mg/dL on a single test or a series of recent tests averaging .50 mg/dL).6 In our more recent data (2000-2009; n 5 49,000 tests), blood test results of 40-,51 and .51 mg/dL represented 4.3% and 2.1% of tests, respectively. Given that repeated intermittent high exposures can lead to a high body burden over time, 6
which may lead to kidney damage, the precautionary principle would suggest that the present OSHA standard, even if consistently enforced, may not be sufficiently protective. Weaknesses in our data include lack of a complete work history, lack of data for possible confounders (eg, hypertension and diabetes), missing data for race for most of the cohort, and possibly inaccurate matching with USRDS and NDI registries due to the lack of social security number information in most of our cohort. However, previous work by Williams et al42 has shown that with name and date of birth, investigators can attain 92% sensitivity and 92% specificity using NDI, a result also likely to be true for USRDS. We have studied a large cohort with occupational exposure to lead; lead is a relatively common exposure among adult workers. We found evidence of increased ESRD incidence at high blood levels, primarily among nonwhites. Data are limited by the lack of detailed work history and reliance on only a few blood lead tests to estimate level of exposure. Strengths of our study include a large cohort, documented blood lead levels, and use of ESRD incidence, rather than reliance on chronic kidney disease mortality. This is a young cohort, and further follow-up is needed to confirm our findings.
ACKNOWLEDGEMENTS We acknowledge the following people for help providing data access and matching: Susan Payne (California Department of Health Services), Beth Forrest (USRDS), Michele Goodier (NDI); ABLES Program: Walter Alarcon (NIOSH), Kenneth D. Rosenman (Michigan State University), Thomas St. Louis (Connecticut Department of Health), Henry A. Anderson, MD (Wisconsin Department of Health and Family Services), Jay Devasundaram (Pennsylvania Department of Health), Ed Socie, PhD (Ohio Department of Health), David J. Valiante (New Jersey Department of Health and Senior Services), Alicia Fletcher (New York Department of Health), Rita Gergely (Iowa Department of Health), Erik Zabel (Minnesota Department Health), and Bob Nicotera (Massachusetts Department of Labor). Support: This work was supported by NIOSH award R01OH00898. Financial Disclosure: The authors declare that they have no other relevant financial interests.
SUPPLEMENTARY MATERIAL Item S1: Race imputation methods. Note: The supplementary material accompanying this article (http://dx.doi.org/10.1053/j.ajkd.2013.12.005) is available at www.ajkd.org
REFERENCES 1. US Environmental Protection Agency. Air Quality Criteria for Lead (Final). EPA-600/8-77/017. Research Triangle Park, NC: National Center for Environmental Assessment-RTP Office, Office of Research and Development, US Environmental Protection Agency; 1977. Am J Kidney Dis. 2014;-(-):---
Lead Exposure and Incident ESRD 2. Pirkle JL, Brody DJ, Gunter EW, et al. The decline in blood lead levels in the United States. The National Health and Nutrition Examination Surveys (NHANES). JAMA. 1994;272(4):284-291. 3. US Environmental Protection Agency. Outdoor air. Exhibit 2.5. Lead emissions in the US by source category 1970-2005. http://www. epa.gov/ncea/roe/index.htm. Accessed January 6, 2014. 4. Staudinger KC, Roth VS. Occupational lead poisoning. Am Fam Physician. 1998;57(4):719-726. 731-712. 5. Centers for Disease Control and Prevention. Blood lead level data, 2002-2005. http://www.cdc.gov/niosh/topics/ABLES/pdfs/2 002-2005lead_data.pdf. Accessed January 6, 2014. 6. Occupational Safety and Health Administration. Lead standards. https://www.osha.gov/SLTC/lead. Accessed January 6, 2014. 7. Levey AS, Stevens LA, Schmid CH, et al. CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration). A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604-612. 8. Collins AJ, Foley RN, Herzog C, et al. US Renal Data System 2012 annual data report. Am J Kidney Dis. 2013;61(1)(suppl 1):e1-e480. 9. US Environmental Protection Agency. Air Quality Criteria for Lead (2006) Final Report. EPA/600/R-05/144aF-bF. Research Triangle Park, NC: National Center for Environmental AssessmentRTP Office, Office of Research and Development; 2006. 10. Ekong E, Jaar B, Weaver V. Lead-related nephrotoxicity: a review of the epidemiologic evidence. Kidney Int. 2006;70(12): 2074-2084. 11. Evans M, Elinder CG. Chronic renal failure from lead: myth or evidence-based fact? Kidney Int. 2011;79(3):272-279. 12. Cocco P, Hua F, Boffetta P, et al. Mortality of Italian lead smelter workers. Scand J Work Environ Health. 1997;23(1):15-23. 13. Steenland NK, Thun MJ, Ferguson CW, Port FK. Occupational and other exposures associated with male end-stage renal disease: a case/control study. Am J Public Health. 1990;80(2):153-157. 14. Cooper W, Wong O, Kheifets L. Mortality among employees of lead battery plants and lead-producing plants, 19471980. Scand J Work Environ Health. 1985;11(5):331-345. 15. Steenland K, Selevan S, Landrigan P. The mortality of lead smelter workers: an update. Am J Public Health. 1992;82(12): 1641-1644. 16. Staessen JA, Lauwerys RR, Buchet JP, et al. Impairment of renal function with increasing blood lead concentrations in the general population. The Cadmibel Study Group. N Engl J Med. 1992;327(3):151-156. 17. Payton M, Hu H, Sparrow D, Weiss ST. Low-level lead exposure and renal function in the Normative Aging Study. Am J Epidemiol. 1994;140(9):821-829. 18. Chia KS, Jeyaratnam J, Tan C, Ong HY, Ong CN, Lee E. Glomerular function of lead-exposed workers. Toxicol Lett. 1995;77(1-3):319-328. 19. Kim R, Rotnitsky A, Sparrow D, Weiss S, Wager C, Hu H. A longitudinal study of low-level lead exposure and impairment of renal function. The Normative Aging Study. JAMA. 1996;275(15): 1177-1181. 20. Lin JL, Tan DT, Hsu KH, Yu CC. Environmental lead exposure and progressive renal insufficiency. Arch Intern Med. 2001;161(2):264-271. 21. Lin JL, Yu CC, Lin-Tan DT, Ho HH. Lead chelation therapy and urate excretion in patients with chronic renal diseases and gout. Kidney Int. 2001;60(1):266-271. 22. Weaver VM, Lee BK, Ahn KD, et al. Associations of lead biomarkers with renal function in Korean lead workers. Occup Environ Med. 2003;60(8):551-562.
Am J Kidney Dis. 2014;-(-):---
23. Tsaih SW, Korrick S, Schwartz J, et al. Lead, diabetes, hypertension, and renal function: the Normative Aging Study. Environ Health Perspect. 2004;112(11):1178-1182. 24. Lin JL, Lin-Tan DT, Li YJ, Chen KH, Huang YL. Lowlevel environmental exposure to lead and progressive chronic kidney diseases. Am J Med. 2006;119(8):707 e701-709. 25. Weaver VM, Griswold M, Todd AC, et al. Longitudinal associations between lead dose and renal function in lead workers. Environ Res. 2009;109(1):101-107. 26. Navas-Acien A, Tellez-Plaza M, Guallar E, et al. Blood cadmium and lead and chronic kidney disease in US adults: a joint analysis. Am J Epidemiol. 2009;170(9):1156-1164. 27. Roscoe RJ, Ball W, Curran JJ, et al. Adult blood lead epidemiology and surveillance—United States, 1998-2001. MMWR Surveill Summ. 2002;51(11):1-10. 28. Centers for Disease Control and Prevention. Adult blood lead epidemiology and surveillance—United States, 2008-2009. MMWR Morbid Mortal Wk Rep. 2011;60(25):841-845. 29. USRDS. Completeness and reliability of USRDS data: comparisons with the Michigan Kidney Registry. Am J Kidney Dis. 1992;20(5)(suppl 2):84-88. 30. Calvert GM, Steenland K, Palu S. End-stage renal disease among silica-exposed gold miners. A new method for assessing incidence among epidemiologic cohorts. JAMA. 1997;277(15): 1219-1223. 31. Steenland K, Sanderson W, Calvert GM. Kidney disease and arthritis in a cohort study of workers exposed to silica. Epidemiology. 2001;12(4):405-412. 32. Radican L, Wartenberg D, Rhoads GG, et al. A retrospective occupational cohort study of end-stage renal disease in aircraft workers exposed to trichloroethylene and other hydrocarbons. J Occup Environ Med. 2006;48(1):1-12. 33. US Renal Data System. ADR reference tables. http://www. usrds.org/reference.aspx. Accessed June 9, 2013. 34. Robinson CF, Schnorr TM, Cassinelli RT II, et al. Tenth revision U.S. mortality rates for use with the NIOSH Life Table Analysis System. J Occup Environ Med. 2006;48(7):662-667. 35. Schubauer-Berigan MK, Hein MJ, Raudabaugh WM, et al. Update of the NIOSH life table analysis system: a person-years analysis program for the windows computing environment. Am J Ind Med. 2011;54(12):915-924. 36. Fadrowski JJ, Abraham AG, Navas-Acien A, Guallar E, Weaver VM, Furth SL. Blood lead level and measured glomerular filtration rate in children with chronic kidney disease. Environ Health Perspect. 2013;121(8):965-970. 37. Kosnett MJ, Wedeen RP, Rothenberg SJ, et al. Recommendations for medical management of adult lead exposure. Environ Health Perspect. 2007;115(3):463-471. 38. Schwartz BS, Hu H. Adult lead exposure: time for change. Environ Health Perspect. 2007;115(3):451-454. 39. Spivey A. The weight of lead. Effects add up in adults. Environ Health Perspect. 2007;115(1):A30-A36. 40. Checkoway HPN, Kriebel D. Research Methods in Occupational Epidemiology. New York, NY: Oxford University Press; 2004. 41. Evans M, Fored CM, Nise G, Bellocco R, Nyrén O, Elinder CG. Occupational lead exposure and severe CKD: a population-based case-control and prospective observational cohort study in Sweden. Am J Kidney Dis. 2010;55(3):497-506. 42. Williams BC, Demitrack LB, Fries BE. The accuracy of the National Death Index when personal identifiers other than Social Security number are used. Am J Public Health. 1992;82(8): 1145-1147.
7
