ADVANCEMENT OF THE SCIENCE
nificantly higher odds of having a BLL >0.76 μg/dl compared with individuals not in pov- erty, after controlling for various demographic variables. These odds ratios were noteworthily high, with individuals in and near poverty showing almost twice the odds of a BLL >0.76 μg/dl. Subsequent age stratification of this third logistic regression model revealed di er- ences in the association’s strength. That is, for individuals living near poverty, the association was strongest among those individuals in the age group 35–44 years. For individuals living in poverty, this association was strongest for both the age groups 35–44 years and 45–54 years. These odds ratios represent drastically increased odds of a BLL >0.76 μg/dl. Additionally, the change in total log-odds estimates revealed significant inverse associa- tions between the poverty-income ratio and log-odds of a BLL >0.76 ug/dL. The signifi- cant estimates for models 2 and 3 represent that as income increases, total log-odds of a BLL >0.76 μg/dl decreases. Age stratification also revealed that the relationship between this change in poverty-income ratio and log- odds of a higher BLL was significant only in the age groups 35–44 years and 45–54 years, which was true for age-stratified modeling of BLLs and categorical poverty status. This study represents an important analysis in a nationally representative sample of the demographic factors that play into an indi- vidual’s risk of higher BLLs. It is clear from the presented findings that individuals living near or in poverty do, indeed, have signifi- cantly increased odds of increased BLLs, after adjusting for various other factors. The con- sideration of race, age, gender, and lifetime cigarette use in these models further supple- ments the significance of these findings. These findings align with existing research on lead exposure, in that demographic factors con- tribute greatly to BLLs (Bernard & McGeehin, 2003; Morales et al., 2005; Yeter et al., 2020). These studies examine threshold lead levels of 5 μg/dl; however, their consideration of vari- ous demographic and socioeconomic factors reflect the findings of this present analysis. Relevance This study’s quantification of the odds of hav- ing BLLs above the sample median provides valuable insight to public and environmental health practitioners. While great strides have been made in equal e orts of lead remediation
TABLE 4
Logistic Regression Results for Blood Lead Levels Greater Than 0.76 μg/dl by Poverty Status From the 2017–2018 National Health and Nutrition Examination Survey
Status
Model 1
Model 2 a
Model 3 b
In poverty
Parameter ( SE )
-0.0151 (0.1340)
0.6794 (0.1293)
0.6009 (0.1504)
p -value
.9120
<.0001
.0012
OR [95% CI]
0.985 [0.740, 1.311]
1.973 [1.497, 2.599]
1.824 [1.324, 2.513]
Near poverty
Parameter ( SE )
0.1463 (0.1168)
0.6553 (0.0806)
0.5560 (0.1069)
p -value
.2294
<.0001
.0001
OR [95% CI]
1.158 [0.903, 1.485]
1.926 [1.622, 2.287]
1.744 [1.388, 2.190]
Not in poverty
Parameter ( SE )
ref. ref. ref.
ref. ref. ref.
ref. ref. ref.
p -value
OR [95% CI]
a Model 2 was adjusted for race, gender, and age. b Model 3 was additionally adjusted for history of smoking ≥100 cigarettes. Note. CI = confidence interval; ref. = reference value.
in the U.S., this analysis reveals there are still clear disparities. After adjustment for demo- graphic factors, individuals living in or near poverty remain the group with the greatest odds of existing in the above-the-median group for BLLs. The significant variations in prevalence of above-the-median BLLs across demographic groups reflect that individuals of lower socio- economic status and minority groups continue to be in need of focused interventions. Limitations This study is subject to limitations. As is ever- present with survey research, the reliance on self-reporting of demographic characteristics can result in under- or over-reporting. Also, NHANES is not administered to individuals who are institutionalized or incarcerated, which limits study generalizability. Further, these data are cross-sectional, which makes it impossible to draw conclusions based on temporality. Moreover, the use of an arbitrary BLL threshold of 0.76 μg/dl, while producing significant results, does not consider lower levels at which lead is known to negatively a ect one’s health. The exclusion of individuals from 6–18 years from the stratified modeling also lim-
its generalizability and, because younger people are at a greater risk of health issues from lead exposure (Cui et al., 2022; Hou et al, 2013), this study omits an important subset of the population. It is also possible that residual confounding is present due to nonconsideration of living arrangements that are surveyed in the NHANES data. Yeter et al. (2020) also report a significant associa- tion between health insurance coverage and BLL, which represents just one of the addi- tional adjustment variables to consider in future analyses. The use of the 100-cigarette history as a smoking classification might omit information that could better adjust the odds of having a BLL greater than the sample median. Strengths This study draws from NHANES data which, when analyzed with proper consideration of weights, strata, and clustering, is a nation- ally representative sample of noninstitution- alized, nonincarcerated people residing in the U.S. Given this foundation, this study is highly generalizable to people in the U.S. who are >18 years, excluding the previously mentioned groups. The consideration of vari-
12
Volume 86 • Number 9
Powered by FlippingBook