Blood Pressure in Relation to Environmental Lead Exposure in the National Health and Nutrition Examination Survey 2003 to 2010Novelty and Significance
In view of the declining environmental lead exposure in the United States, we analyzed the National Health and Nutrition Examination Survey (2003–2010) for association of blood pressure and hypertension with blood lead. The 12 725 participants included 21.1% blacks, 20.5% Hispanics, 58.4% whites, and 48.7% women. Blacks compared with non-Blacks had higher systolic and diastolic pressures (126.5 versus 123.9 and 71.9 versus 69.6 mm Hg) and higher hypertension prevalence (44.7 versus 36.8%). Blood lead was lower in whites than in non-whites (1.46 versus 1.57 μg/dL) and in women than in men (1.25 versus 1.80 μg/dL). In multivariable analyses of all participants, blood lead doubling was associated with higher (P≤0.0007) systolic and diastolic pressure (+0.76 mm Hg; 95% confidence interval, 0.38–1.13 and +0.43 mm Hg; 0.18–0.68), but not with the odds of hypertension (0.95; 0.90–1.01; P=0.11). Associations with blood lead were nonsignificant (P≥0.09) for systolic pressure in women and for diastolic pressure in non-whites. Among men, systolic pressure increased with blood lead (P≤0.060) with effect sizes associated with blood lead doubling ranging from +0.65 mm Hg in whites to +1.61 mm Hg in blacks. For systolic pressure, interactions of ethnicity and sex with blood lead were all significant (P≤0.019). In conclusion, small and inconsistent effect sizes in the associations of blood pressure with blood lead likely exclude current environmental lead exposure as a major hypertension cause in the United States.
Voluntary and regulatory restrictions since the 1970s on the use of lead in gasoline, paint, and soldered food cans resulted in a progressive decline in the exposure of populations to lead. In the United States, the National Health and Nutrition Examination Survey (NHANES) documented a progressive decline in the geometric blood lead concentration over time. Among adults, mean blood lead levels decreased from 13.1 μg/dL in NHANES II (1976–1980)1,2 to 2.76 μg/dL in NHANES III (1988–1994)3 and next to 1.64 μg/dL in NHANES IV (1999–2002).3
High-level lead exposure causes hypertension.4,5 In a previous meta-analysis, we reported that a 2-fold increase in the blood lead concentration was associated with a higher blood pressure with estimated effect sizes of 1.0 mm Hg (95% confidence interval [CI], 0.5–1.4 mm Hg) systolic and 0.6 mm Hg (0.4–0.8 mm Hg) diastolic.4 In our analysis of NHANES III data,6 the median blood lead concentrations among blacks and whites of either sex ranged from 2.1 to 4.2 μg/dL. The multivariable-adjusted changes in blood pressure associated with a doubling of blood lead were inconsistent across the 4 ethnicity-sex strata and ranged from +0.1 mm Hg (P=0.80) to +1.2 mm Hg (P=0.004) systolic and from −0.6 mm Hg (P=0.0003) to +0.5 mm Hg (P=0.047) diastolic.6 In view of the continuing declining environmental lead exposure in the United States,1,3 and the apparent ethnic diversity in NHANES III in the associations of blood pressure with blood lead,6,7 we aimed at reanalyzing the relationship between blood pressure and blood lead in NHANES IV (2003–2010).
The US National Center for Health Statistics (Centers for Disease Control and Prevention, Atlanta, GA) conducted NHANES IV (1999–2012). Interviews were conducted at the participants’ homes. The comprehensive physical examinations, which included measurements of anthropometric characteristics, blood pressure, and collection of blood and urine samples, took place at mobile examination centers. The National Center for Health Statistics Institutional Review Board approved the interviews, physical examinations, and the procedure to obtain written informed consent. The details of the field work are described in the Expanded Methods in the online-only Data Supplement (page S2).
Selection of Participants
The NHANES IV data considered for the current analysis were collected in 7 stages: 1999 to 2000, 2001 to 2002, 2003 to 2004, 2005 to 2006, 2007 to 2008, 2009 to 2010, and 2011 to 2012. The pooled data initially comprised 71 916 people examined from 1999 until 2012. In keeping with our previous report,6 we planned to account for dietary habits. This forced us to excluded participants examined from 1999 to 2002, because the protocol for collecting dietary information substantially changed in 2002 and people examined in 2011 and 2012, because dietary information was lacking in the online database (accessed February 20, 2014). After exclusion of participants with missing or unreliable data, the number of analyzed participants totaled 12 725. The Expanded Methods in the online-only Data Supplement (pages S2–S3) provide detailed information on selection and exclusion criteria and on the representativeness of the participants retained in the analysis. Self-reported ethnicity was categorized as non-Hispanic white, non-Hispanic black, and Hispanic.
At mobile examination centers, trained observers measured anthropometric characteristics and blood pressure. Blood pressure was the average of ≤3 readings. The number of blood pressure readings available for analysis was 3 in 11 601 participants (91.2%), 2 in 626 (4.9%), and only 1 in 498 (3.9%). Hypertension was a blood pressure of ≥140 mm Hg systolic or 90 mm Hg diastolic or the use of antihypertensive drugs. Pulse pressure was the difference of systolic minus diastolic blood pressure. Mean arterial pressure was diastolic blood pressure plus one third of pulse pressure. The Expanded Methods in the online-only Data Supplement (pages S3–S5) provide a thorough description of the methods used for administering questionnaires, recording dietary habits, and the methods used for the biochemical measurements including blood lead.
For database management and statistical analysis, we used SAS software, version 9.3 (SAS Institute Inc, Cary, NC). For variables that required a logarithmic transformation to approximate a normal distribution, including blood lead, serum cotinine, and γ-glutamyltransferase, and dietary calcium and caffeine, we reported the central tendency and spread of the data as the geometric mean and the interquartile range. Between-group comparisons of means relied on the standard normal z test or Tukey test for multiple comparisons. For between-group comparisons of proportions, we applied the χ2 statistic test with Bonferroni correction of the significance levels if multiple groups were involved.
Details of the analysis plan and the statistical tests applied appear in the Expanded Methods in the online-only Data Supplement (pages S5–S6). In a first step of the analysis, we plotted mean values of systolic and diastolic blood pressure by deciles of the blood lead concentration for each of the 6 ethnicity-sex strata separately. Next, we searched for covariables significantly and independently associated with blood pressure in a stepwise regression procedure with P values for explanatory variables to enter and stay in the model set at 0.15. After having determined the standard set of covariables to adjust for, we computed for each ethnicity-sex stratum the multivariable-adjusted associations between blood pressure and blood lead. We tested between-group differences in these association by introducing the appropriate interaction terms with blood lead in the models.
Characteristics of Participants by Ethnicity and Sex
The 12 725 participants included 2692 (21.1%) blacks, 2607 (20.5%) Hispanics, 7426 (58.4%) whites, and 6199 (48.7%) women. The characteristics listed in Table 1 are significantly different between the 3 ethnic groups (P<0.0001) except for pulse pressure (P=0.66) and heart rate (P=0.30). The online-only Data Supplement includes a detailed description of the differences in the characteristics of participants by ethnicity (page S6) and sex (page S7). The blood lead concentration was lower (P<0.0001) among whites when compared with black and Hispanic participants (1.46 versus 1.57 μg/dL) with no difference (P=0.12) between Hispanics (1.55 μg/dL) and blacks (1.60 μg/dL). Women had a lower blood lead concentration than men had (1.25 versus 1.80 μg/dL; P<0.0001). When adjusted for hematocrit, the blood lead concentration remained lower in women than in men (1.29 versus 1.74 μg/dL; P<0.0001). Figure S1 in the online-only Data Supplement describes the distributions of blood lead by ethnicity and sex.
The Figure shows that in the unadjusted analysis of the 6 ethnicity-sex strata, systolic blood pressure increased (P<0.0001) with higher blood lead concentration and that a linear model was adequate to describe the data. The corresponding associations with diastolic blood pressure were significantly positive (P≤0.018) or not significant (P≥0.33) but also suggested that a linear model accurately captured the data (Figure S2).
In unadjusted analyses of all participants, systolic and diastolic pressure increased (P<0.0001) with higher blood lead (Table S1). The effect sizes associated with a doubling of the blood lead concentration were 4.80 mm Hg (95% CI, 4.46–5.14) for systolic pressure and 0.67 mm Hg (CI, 0.45–0.89) for diastolic pressure (Table S1). Considering the six ethnicity-sex strata, systolic and diastolic pressure also increased with higher blood lead (P≤0.018), except for diastolic pressure (Table S1) in white women (P=0.69) and non-Black men (P≥0.33). In unadjusted analyses, pulse pressure and mean arterial blood pressure consistently increased (P≤0.0095) with blood lead in all ethnicity-sex strata and in all participants (Table S2, page S7).
Among all participants, 4893 (38.5%) had hypertension. The diagnosis of hypertension rested on a systolic pressure of ≥140 mm Hg in 1903 subjects, a diastolic pressure of ≥90 mm Hg in 228, an elevation of both systolic and diastolic pressure in 406, and the use of antihypertensive drugs in 2356 participants, in whom the aforementioned blood pressure criteria were not met. The unadjusted odds ratio for a doubling of the blood lead concentration (Table S3) was 1.58 (95% CI, 1.52–1.65) in all participants. Across the 6 ethnicity-sex strata, except for Hispanic men (P=0.12), the unadjusted odd ratios were significant (P<0.0001), ranging from 1.51 in Hispanic women to 2.17 in white women (Table S3).
Identification of Covariables
Covariables of systolic and diastolic pressure, as selected by stepwise regression, appear in Table 2. Age was the most important independent covariable, explaining 20.5% and 11.4% of the systolic and diastolic variance, respectively (Table 2). With whites as reference, both blacks and Hispanics had higher systolic pressure (Table 2). Blacks also had higher diastolic pressure (Table 2). Systolic pressure was inversely associated with heart rate, whereas the opposite was the case for diastolic pressure. Both systolic and diastolic pressure significantly (P≤0.0035) and independently increased with body mass index, γ-glutamyltransferase as index of alcohol intake, hematocrit and the dietary sodium:potassium ratio. Users of antihypertensive drugs had higher blood pressure. Diastolic pressure was independently and inversely correlated with serum cotinine. College graduation was associated with lower systolic and diastolic pressure. Taken together, all covariables selected explained 25.7% of systolic pressure and 17.9% of diastolic pressure.
The online-only Data Supplement (Table S4, page S8) shows that the covariables associated in stepwise regression with pulse pressure and mean arterial pressure were similar as those correlated with systolic and diastolic pressure and together explained 35.6% of pulse pressure and 14.8% of mean arterial pressure. On the basis of above results, we adjusted all regression models relating blood pressure components to blood lead for ethnicity and sex (as appropriate), the linear and squared terms of age, body mass index, heart rate, hematocrit, serum total calcium, γ-glutamyltransferase and cotinine, the dietary sodium:potassium intake ratio, attainment of a college grade, and antihypertensive drug treatment.
Multivariable-Adjusted Association of Blood Pressure With Blood Lead
Among women, the multivariable-adjusted associations between systolic and diastolic pressure and blood lead (Table 3) did not reach significance in any of the ethnic groups (P≥0.090), except for diastolic pressure in white women, which was +0.73 mm Hg higher (95% CI, +0.23 to +1.24; P=0.0045) for a 2-fold increase in blood lead. In all women combined, the effect sizes associated with a doubling of the blood lead concentration were +0.58 mm Hg (95% CI, +0.01 to +1.17; P=0.050) for systolic pressure and +0.43 (95% CI, 0.07 to +0.80; P=0.021) for diastolic pressure. Among men, systolic pressure was significantly and independently associated with the blood lead concentration in black and Hispanic men (P≤0.038), whereas the corresponding association in white men was statistically weaker (P=0.060). The effect sizes for a doubling of the blood lead concentration ranged from +1.61 mm Hg in blacks to +0.65 mm Hg in whites. In all men combined, the effect size was +0.79 mm Hg (95% CI, +0.30 to +1.27; P=0.0015). The relationship between diastolic pressure and blood lead did not reach formal significance (P≥0.062) among black and Hispanic men. In white men, diastolic pressure was significantly and positively associated with blood lead (effect size, +0.70 mm Hg; 95% CI, +0.24 to +1.17; P=0.0032). Among all participants, for each 2-fold increase in the blood lead concentration, blood pressure components increased, by +0.76 mm Hg (95% CI, +0.38 to +1.13; P<0.0001) for systolic pressure and by +0.43 mm Hg (95% CI, +0.18 to +0.68; P=0.0007) for diastolic pressure (Table 3).
For systolic blood pressure, all interaction terms of ethnicity and sex with blood lead were significant (P≤0.019). For diastolic pressure, the interaction terms of ethnicity and sex with lead were nonsignificant (P≥0.17), except for the interaction between Hispanic ethnicity and blood lead in relation to diastolic blood pressure. This interaction indicated that for a doubling of blood lead the increase in diastolic pressure among Hispanics was 0.70 mm Hg less (−1.20 to −0.20 mm Hg; P=0.0057) than in the other ethnicities.
The online-only Data Supplement gives full information on the multivariable-adjusted associations of pulse pressure and mean arterial pressure with blood lead in all participants and in the 6 ethnicity-sex strata (Table S5, pages S8–S9). Among all participants, mean arterial pressure was 0.54 mm Hg higher (95% CI, 0.29–0.79; P<0.0001) for each 2-fold increase in the blood lead concentration, whereas there was no significant association between pulse pressure and blood lead even when all participants were pooled (Table S5). For pulse pressure, interactions of ethnicity and sex with blood lead were all significant (P≤0.027). For mean arterial pressure, all interaction terms of ethnicity and sex with blood lead were nonsignificant (P≥0.096).
Multivariable-Adjusted Risk of Hypertension in Relation to Blood Lead
The odds of having hypertension associated with a doubling of the blood lead concentration (Table 4) only reached formal significance in black women (P=0.049) and Hispanic men (P=0.042), in whom the odds ratios were 0.82 (95% CI, 0.67–0.99) and 0.84 (CI, 0.71–0.99), respectively. For all other subgroups, the odds ratios were not statistically significant (P≥0.12). Among all participants, the odds ratio was 0.95 (95% CI, 0.90–1.01; P=0.11).
The sensitivity analyses (Tables S6–S8, pages S9–S10) confirmed the primary findings.
We undertook our current study in view of the steadily declining blood lead levels in the United States. The blood lead concentration averaged 13.1 μg/dL1 in NHANES II (1976–1980), declined to 2.76 μg/dL in NHANES III (1988–1994),3 and to 1.64 μg/dL and 1.41 μg/dL in NHANES IV (1999–20023 and 2005–20068). In our present study (2003–2010), the mean blood lead concentration was 1.51 μg/dL. The key findings can be summarized in 4 points. First, across 6 ethnicity-sex strata, the relationship between the blood pressure components was inconsistent, usually with larger effects sizes among men than women and among blacks than non-blacks. Second, the effect sizes, although significant for systolic, diastolic, and mean arterial pressure in pooled analyses of all women and men combined, were all smaller than 0.79 mm Hg. Third, pulse pressure was not related to blood lead, except for a weak association in Hispanic men. Fourth, the small effect sizes explain why overall the odds of having hypertension was not associated with the blood lead concentration.
A comprehensive review of previous studies of the association of blood pressure or the prevalence of hypertension with lead exposure3,4,8 is beyond the scope of this article but is available in the online-only Data Supplement (pages S10–S12). The current literature shows discrepancy between studies in populations and workers. The explanations that are commonly put forward for this apparent discrepancy are the higher statistical power in large epidemiological surveys relative to smaller occupational cohorts and selection bias, often referred to as the healthy worker effect. Considering the available literature, NHANES is undoubtedly the most appropriate data source to address the issue of the potential association of blood pressure or hypertension with environmental lead exposure. NHANES conducted by the National Center for Health Statistics is the principal resource for tracking progress in reducing lead in the environment. Because NHANES surveyed a large probability sample of the general population, the findings can be generalized to the United States as a whole. Moreover, individuals with high exposure or at risk of high exposure were excluded from the NHANES sample. Previous NHANES phases showed a substantial decline in blood lead levels.1,3,8 The percentage of adults with blood lead level of 10 μg/dL or higher was as low as 0.7% in 1999 to 2002.3
Two previous NHANES IV studies reported on the association of blood pressure or the risk of hypertension with blood lead.3,8 Both studies confirmed that blood lead levels continue to decline among adults in United States, but that racial and ethnic disparities persist.3,8 One study reported that blood lead levels were significantly associated with higher systolic blood pressure among black men and women, but not among white or Mexican-American participants, and that blood lead was significantly correlated with higher diastolic blood pressure among white men and women and black men, whereas a negative association was observed in Mexican-American men.8 The second study reported that blood lead levels remained associated with a higher burden of chronic kidney disease and peripheral arterial disease.3 Our current analysis essentially showed that the low levels of lead that exist today in the United States have little influence on blood pressure. Our analytic approach also differed from the 2 previously published reports in several aspects. First, in unadjusted analyses, we tested the adequacy of the linear model to describe the relationship between blood pressure and blood lead. Second, in our multivariable-adjusted analyses, we analyzed blood lead as a continuous variable instead of examining the linear trend across quartiles3 or contrasting the bottom with the top decile.8 Third, we analyzed both the pulsatile components (systolic pressure and pulse pressure) and the steady components (diastolic pressure and mean arterial pressure) of blood pressure. Our observation that pulse pressure was unrelated to blood lead is at variance with studies relating peripheral arterial disease or cardiovascular disease with lead exposure.9,10 However, smoking was a major confounder in these reports.9,10 Fourth, we adjusted for a large number of covariables, including aspects of lifestyle that previous NHANES studies3,8 did not consider, such as nutritional factors, social status, and serum cotinine and γ-glutamyltransferase, as biomarkers of smoking and drinking alcohol.
Our current study must be interpreted within the context of its limitations. First, the cross-sectional nature of our current analysis does not allow to make any causal interferences about the association between blood pressure and the prevalence of hypertension in relation to lead exposure. Second, we cannot ascertain that we accounted for all confounders, in particular, when we assessed the association between blood pressure and blood lead across 6 ethnicity-sex strata. Third, blood pressure was the average of 3 conventional blood pressure readings. Out-of-the-office blood pressure measurement by ambulatory monitoring or self-measurement at home is the current state of the art to assess blood pressure.11 Fourth, a potential limitation of our study was that we did not measure bone lead as exposure marker. Approximately 95% of the total body burden of lead is present in the skeleton, and measurement of bone lead levels can provide a more reliable measure of the internal dose.12 Some investigators reported that hypertension was associated with bone lead, but not with blood lead.13,14 A possible drawback of using bone lead as exposure index is that bone levels increase with age, as blood pressure does, making it difficult to correct for age effects. Furthermore, the upper end of bone lead levels reported in previous studies13,14 was characteristic of occupational rather than environmental exposure. Blood lead reflects both recent exogenous exposure and endogenous redistribution of the lead stored in bone, but may underestimate the internal dose of lead.
In our current study, the increase from the low to high end of the distribution was ≈10-fold. The 5th to 95th percentile interval encompassed 0.54 and 4.35 μg/dL. In the unlikely event that a patient would move from the bottom to the top of the blood lead distribution, which would translate in a maximum increase in blood pressure by ≈5 mm Hg systolic or ≈3 mm Hg diastolic. Furthermore, in our current cross-sectional analysis, there was no consistent relationship of the blood pressure level or the prevalence of hypertension with the blood lead concentration. The association between hypertension and blood lead was not significant. Accordingly, lead exposure within the range studied might not entail any excess morbidity or mortality attributable to hypertension and its cardiovascular complications.15 However, more prospective population studies with assessment of both fatal and nonfatal cardiovascular end points are required to confirm this assertion. The currently available prospective population studies15–19 are contradictory, reporting positive16–19 or null15 associations between outcome and lead exposure, but to our knowledge only 215,16 accounted for both fatal and nonfatal events. Not having the nonfatal outcomes in an era when invasive cardiologist remove obstructions and restore patency of coronary arteries, when coronary bypass surgery became a low-risk procedure, and when stroke units are delivering specialized intensive care, recording the nonfatal events should become the state of the art. Finally, for now, no study captured the low end of the exposure–response relationship for blood lead levels and end points of cardiovascular function in lead workers. Therefore, studies specifically addressing these issues in a longitudinal follow-up of lead exposed workers who will go from no previous occupational (general population blood lead levels) to occupational lead exposures are needed.
Drs Ryman and Boreiko, International Lead Zinc Research Organization (Triangle Park, NC), facilitated access to the National Health and Nutrition Examination Survey (NHANES) IV data. Annick De Soete provided expert clerical assistance.
Sources of Funding
The Japan Society for the Promotion of Science Postdoctoral Fellowships for Research Abroad supported Dr. Hara’s fellowship in Leuven. The European Union (grant FP7-HEALTH-2007-A-201550), HYPERGENES, and InterOmics (PB05 MIUR-CNR Italian Flagship Project) provided financial support for the genotyping studies. The European Union (grants IC15-CT98-0329-EPOGH, LSHM-CT-2006-037093-InGenious HyperCare, HEALTH-2007-2.1.1-2-HyperGenes, HEALTH-2011.2.4.2-2-EU-MASCARA, HEALTH-F7-305507 HOMAGE, and the European Research Council Advanced Researcher Grant-2011-294713-EPLORE) gave support to the Studies Coordinating Centre, Leuven, Belgium. The Fonds voor Wetenschappelijk Onderzoek Vlaanderen, Ministry of the Flemish Community, Brussels, Belgium (G.0881.13 and G.0880.13) also supported the Flemish Study on Environment, Genes and Health Outcomes (FLEMENGHO) study. The International Lead and Zinc Organization (Durham, NC) provided a nonbinding grant to the Studies Coordinating Centre for a cohort study in lead exposed workers. The funding sources had no role in study design, data extraction, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all of the data in the study and had the responsibility for the decision to submit for publication.
This article was sent to Robert M. Carey, Consulting Editor, for review by expert referees, editorial decision, and final disposition.
The online-only Data Supplement is available with this article at http://hyper.ahajournals.org/lookup/suppl/doi:10.1161/HYPERTENSIONAHA.114.04023/-/DC1.
- © 2014 American Heart Association, Inc.
- Vupputuri S,
- He J,
- Muntner P,
- Bazzano LA,
- Whelton PK,
- Batuman V
- Agarwal S,
- Zaman T,
- Tuzcu EM,
- Kapadia SR
- O’Brien E,
- Asmar R,
- Beilin L,
- Imai Y,
- Mancia G,
- Mengden T,
- Myers M,
- Padfield P,
- Palatini P,
- Parati G,
- Pickering T,
- Redon J,
- Staessen J,
- Stergiou G,
- Verdecchia P
- Martin D,
- Glass TA,
- Bandeen-Roche K,
- Todd AC,
- Shi W,
- Schwartz BS
- Møller L,
- Kristensen TS
- Menke A,
- Muntner P,
- Batuman V,
- Silbergeld EK,
- Guallar E
- Weisskopf MG,
- Jain N,
- Nie H,
- Sparrow D,
- Vokonas P,
- Schwartz J,
- Hu H
Novelty and Significance
What Is New?
In the United States, the National Health and Nutrition Examination Survey documented a progressive decline in blood lead. National Health and Nutrition Examination Survey III (1988–1994) documented ethnic diversity in the association of blood pressure with blood lead. Therefore, we analyzed 12 725 people included in the National Health and Nutrition Examination Survey IV (2003–2010) database.
What Is Relevant?
The association between blood pressure components and blood lead was inconsistent across 6 strata based on ethnicity (blacks, Hispanics, and whites) and sex.
Among 6199 women, the adjusted effect sizes associated with a doubling of blood lead were +0.58 mm Hg (P=0.05) systolic and +0.43 (P=0.021) diastolic.
Among 6526 men, the corresponding effect sizes were +0.79 mm Hg (P=0.0015) and +0.47 (P=0.0072), respectively.
Among all participants, the odds ratio for hypertension associated with a doubling of blood lead was 0.95 (P=0.11).
At the currently declining exposure levels, associations of blood pressure with blood lead were inconsistent across the ethnicity-sex strata. The effect sizes, although significant in pooled analyses of all women and men, were smaller than 0.79 mm Hg and likely exclude current environmental lead exposure in the United States as a clinically meaningful cause of hypertension.