BLOOD
LEAD AND MORTALITY IN THE NATIONAL HEALTH AND NUTRITION EXAMINATION SURVEY II
(NHANES II) COHORT
Mark
Lustberg; Ellen Silbergeld, PhD
9-34
MSTF, University of Maryland, Baltimore, MD 21201; mlustber@umaryland.edu
Abstract
Lead exposure in the environment is ubiquitous, yet
lead has no known biologic value. Several studies, mostly occupational, have
documented potentially harmful consequences of lead exposure in adults –
ranging from carcinogenesis to increasing blood pressure. It is reasonable and relevant then to
consider the relation of lead exposure and mortality in the general population,
for which we used data from the National Health and Nutrition Examination
Survey II (NHANES II) Mortality Survey.
In this cohort, for persons with blood lead levels in the range of 0-29
mcg/dL there appears to be a relation
of mortality and lead level, especially mortality from lung cancer and
respiratory disease.
Introduction
Lead exposure is ubiquitous, yet lead itself is not biologically
useful. Studies indicate that lead
exposure may have diverse toxic effects – ranging from interfering with
cognitive function, to decreasing fertility, to (depending upon dose and age)
increasing blood pressure, to being a carcinogen. There are relatively few studies on the association between lead
exposure and cause-specific mortality, especially in non-occupationally exposed
persons. Accordingly, it is reasonable
and relevant to consider the relation of lead exposure and mortality in the
general population, for which we used recently published data from the National
Health and Nutrition Examination Survey II (NHANES II) Mortality Study.
Methods
The National Center for Health Statistics (NCHS)
conducted The National Health and Nutrition Examination Survey II (NHANES II)
from 1976-1980. A total of 20,322
people aged 12-74 were examined and interviewed for the survey. Recently NCHS has released mortality
follow-up information for individuals aged 30-74 at examination in NHANES II (n
= 9,252). The National Death Index and
the Social Security Administration were used to ascertain vital status of this
group as of 1992. Definite
ascertainment of vital status was nearly complete at >99.9% (n = 9,250).
Individuals (n=2) whose status was unknown were excluded from the analysis. Over the course of follow-up 2,145
individuals died (mortality rate = 23%).
Baseline blood lead was evaluated in
4,292 individuals aged 30-74 (46.4%).
Individuals with missing baseline blood lead levels, and those for whom
no measurements were taken were excluded from the analysis. Individuals with extreme lead values (lead >
30 mcg/dL) were also excluded from the analysis. Of the remaining 4190 individuals, 929 died over follow-up (unweighted mortality rate = 22%)
All analyses were conducted in SUDAAN (Research
Triangle Institute), using the blood lead weights. NCHS provided these weights in order to make the population
surveyed in NHANES II nationally representative.
In general, two sets of regression
models were evaluated, unadjusted and adjusted models. Adjusted models were adjusted for age
(represented as a continuous variable), male sex, race (white, black, other),
education (report any versus no college education), income (report < 20th
percentile income), smoking (quit and cigarettes smoked per day), and location
(urban, rural, and suburban).
Results
After weighting, the cohort (n=4,190) is roughly
representative of the US population as a whole. The cohort has a slight predominance of women (52.4%). Blacks are proportionately represented at
9.7%. Almost a third of the cohort
reported that they were currently smokers, with 12.4% of individuals smoking
less than a pack a day, and 24.0% smoking a pack or more. Almost a third of individuals (29.1%)
reported that they entered college, while 42% reported graduating high school
and not entering college.
Table 1 presents the results of proportional hazards
modeling of the lead-blood pressure relation.
Before adjustment for potential confounders there is a statistically
significant and impressive relation of blood lead and mortality. For a 10 mcg/dL increase in blood lead, the
relative hazard (the hazard ratio, HR) for mortality is 1.58 (CI = 1.35 –
1.86). This relation was robust to
adjustments for age (which was taken as a continuous variable), sex,
urban/suburban/rural location, low income (< 20th percentile),
college education, smoking (none, < 1 pack, 1 pack or more) and race (black,
white, other). After multiple
adjustments a 10 mcg/dL increase in blood lead corresponded to HR = 1.29 (CI =
1.06 -1.58, p=0.02) for mortality.
To examine whether or not the
relation of lead and death varied by cause of death, we divided death into six
categories based on ICD-9 code: cardio-cerebrovascular (40.5 % of deaths) (ICD9
codes 390-459), lung cancer (9.6 % of deaths) (ICD9 code 162), non-lung cancer
(20.9 %) (ICD9 codes 140-240, excluding 162), respiratory disease (10.4 %)
(ICD9 codes 460-519), suicide-accident-homicide (3.4 % of deaths) (ICD9 codes
E800 and above) and other (15.2 % of deaths) (all other ICD9 codes).
Table 1 presents hazard ratios for
mortality per 10 mcg/dL increase in blood lead, by the above categories of
mortality. After adjustment for
potential confounders, deaths from respiratory disease (HR = 1.65, CI=1.03-2.65)
and suicide-accident-homicide (HR = 1.60, CI=1.01-2.56) were significantly (p
< 0.05) associated with blood lead level.
After adjustment for potential confounders, deaths from lung cancer (HR=
1.75, CI=0.89-3.49) were associated with blood lead level at a level of
borderline significance (p=0.10).
These
data indicate a statistically significant relation of lead and death in the
NHANES II cohort. This relation is
robust to adjustments for age, sex, race, smoking, college education, low
income, and urban/rural/suburban location.
The data indicate that after adjustment for potential confounders, over
the blood lead level range of 0-29 mcg/dL, on average mortality increases by
29% per 10 mcg/dL increase in blood lead.
It
is difficult to know if the relation of blood lead and mortality is a product
of uncontrolled (perhaps unknown)
confounding or whether it represents a true biologic association. Some studies indicate that blood lead is a
potential carcinogen [Waalkes, 1995], and other studies show that it increases
blood pressure (which should predispose towards cardiovascular mortality)
[Schwartz, 1995]; though these findings have not been consistently found across
studies.
Interestingly,
our data indicate the strongest associations of blood lead and lung cancer,
respiratory disease, and suicide-accident-homicide mortality; associations
which persist after multiple adjustments.
As lead level is associated with smoking and socioeconomic status, it is
certainly conceivable that the association of lead with the above forms of
mortality represents residual confounding by smoking and socioeconomic
status. Alternatively, it may be that
lead exposure in smokers is one of the causal factors for lung cancer and respiratory
disease, and that lead exposure in individuals with lower socio-economic status
is partly responsible for the increased suicide-homicide-accident mortality in
this group. Two occupational studies on
lead exposure have documented increased lung cancer in lead exposed workers
[Cooper, 1985; Gerhardsson, 1986]; although other occupational studies have
presented conflicting findings [Malcolm, 1982; Dingwall-Fordyce, 1963]. Recent studies by Needleman, et al. have
found associations between lead exposure and delinquent behavior, and lead exposure
and antisocial behavior [Needleman, 1996].
In the end, if lead is a weak
carcinogen or exerts a comparatively small effect on blood pressure, it may be
difficult to demonstrate epidemiologically, especially for low levels of lead
exposure. This is of concern because
blood lead levels have fallen tremendously in the United States since NHANES II
was conducted. Conducted from
1976-1980, the NHANES II cohort median blood lead level was 13 mcg/dL for
individuals aged 30-75. Conducted from
1988-1994, the NHANES III median blood lead level was 3.0 mcg/dL for
individuals aged 30-75 [DHHS, 1996].
Thus,
it could be argued that the range of blood lead concentrations most relevant
today are not adequately considered by the NHANES II data. Less than 10% of individuals in the NHANES
II cohort have blood lead < 3.0 mcg/dL. Accordingly, these data do not speak as to the relation of
mortality (in general, or for any specific cause) for the low blood lead levels
most commonly encountered in the population today.
Perhaps, these data are best viewed
as reinforcing the importance of maintaining low blood lead levels in the
population, possibly offering support to lower the occupational blood lead
standard, which is currently 40 mcg/dL [Landrigan, 1991]. If the relation of blood lead and mortality
in this cohort is real, even a small association of mortality and blood lead
level could be of public health importance.
References
1. Waalkes M (1995), In:
Metal Toxicology. (R Goyer, C Klaassen, M Waalkes, Editors), San Diego, Academic Press, Inc., pp 77-98.
2. Schwartz
J (1995), Archives of Environ. Health; 50:31-37.
3. Cooper
W, et al. (1985), Scandinavian J. of Work Environ. Health; 11:331-345.
4. Gerhardsson
L, et al. (1986), British J. of Industrial Medicine; 43:707-712.
5. Malcolm
D, Barnett H (1982), British J. of Industrial Medicine; 39:404-410.
6.
Dingwall-Fordyce
I, Lane R (1963), British J. of Industrial Medicine; 20:313-315.
7.
DHHS,
National Center for Health Statistics (1996).
Third National Health and Nutrition Examination Survey, 1988-1994,
NHANES III Data Files. (CD-ROM).
8.
Needeleman
H, et al. (1996), JAMA; 275: 363-369.
Table 1: Hazard ratio for death by 10 mcg/dL increase in blood lead level
for participants in the NHANES Mortality Study, results of survival analysis
regression modeling
Model
|
Hazard Ratio |
CI
|
p-value |
|
|
|
|
|
|
Any mortality |
|
|
|
|
Unadjusted |
1.58 |
1.35 - 1.86 |
< 0.01 |
|
Adjusted |
1.29 |
1.06 - 1.58 |
0.01 |
|
|
|
|
|
|
Cardio-Cerebrovascular |
|
|
|
|
Unadjusted |
1.38 |
1.12 - 1.70 |
< 0.01 |
|
Adjusted |
1.13 |
0.84 - 1.53 |
0.40 |
|
|
|
|
|
|
Lung Cancer |
|
|
|
|
Unadjusted |
2.95 |
1.70 - 5.10 |
< 0.01 |
|
Adjusted |
1.75 |
0.89 - 3.49 |
0.10 |
|
|
|
|
|
|
Non-Lung cancer |
|
|
|
|
Unadjusted |
1.37 |
0.93 - 2.02 |
0.11 |
|
Adjusted |
1.26 |
0.81 - 1.99 |
0.30 |
|
|
|
|
|
|
Respiratory Disease |
|
|
|
|
Unadjusted |
1.98 |
1.28 - 3.05 |
< 0.01 |
|
Adjusted |
1.65 |
1.03 - 2.65 |
0.04 |
|
|
|
|
|
|
Suicide/Homicide/Accident |
|
|
|
|
Unadjusted |
2.14 |
1.07 - 4.28 |
0.03 |
|
Adjusted |
1.60 |
1.01 - 2.56 |
0.04 |
|
|
|
|
|
|
Other |
|
|
|
|
Unadjusted |
1.47 |
1.01 - 2.16 |
0.05 |
|
Adjusted |
1.29 |
0.87 - 1.91 |
0.20 |