ADVANCEMENT OF THE SCIENCE
from continuous age, following the standard 8-group classification. BLL data for indi- viduals ≤5 years were not publicly available; therefore, these individuals were excluded from the analysis. Analysis Sample weights, strata, and clusters were pro- vided by NHANES to account for the com- plex sampling design of this survey. Analysis was performed in SAS version 9.4, with con- sideration of complex sampling in all calcula- tions and regression analyses. Weighted population descriptives were cal- culated for poverty groups, cigarette history groups, and other demographic classifications. Additionally, both adjusted and unadjusted logistic regression models were performed to model the odds of a BLL greater than the sam- ple median. Covariates for adjustment were determined based on significant associations and review of existing literature on the subject. Dierences in prevalence of BLL >0.76 μg/dl were assessed for significance using Wald chi- square tests, with p -values <.0500 considered to be significant. Model 1 was unadjusted and simply modeled the odds of a BLL greater than the sample median based on poverty status. Model 2 adjusted for race and ethnicity, gen- der, and age. Model 3 included the 100-ciga- rette history as a covariate. Parameter estimates for these models were considered significant if the Wald chi-square p -value was <.0500. Age-stratified analysis was also performed for the three regression models using the categorical age group vari- able. Parameter estimates for these models were again considered significant at Wald chi-square p -values <.0500. These models were again used to calculate estimates of the change in total log-odds of a BLL >0.76 μg/dl per one unit increase in poverty-income ratio. Age stratification of the fully adjusted model was also performed to derive the change in total log-odds estimate for each age group. For both, p -values <.0500 were considered to be significant. Results Unweighted and weighted counts and per- centages with 95% confidence intervals (CI) by demographic group of the population are displayed in Table 1. Mean BLLs were calcu- lated for each demographic group (Table 2). Individuals >80 years, individuals identify-
TABLE 2
Mean Blood Lead Level by Age, Gender, Race, and Poverty Status From the 2017–2018 National Health and Nutrition Examination Survey
Demographic
Blood Lead Level (μg/dl)
Mean
SE
Minimum, Maximum
Age (years) <18
0.523 0.664 0.806 1.012 1.055 1.330 1.429 1.732 1.087 0.736 0.994 1.047 1.176
0.017 0.045 0.038 0.100 0.052 0.046 0.041 0.075 0.143 0.029 0.031 0.023 0.050
0.05, 6.13
18–24 25–34 35–44 45–54 55–64 65–80
0.05, 17.02
0.09, 8.82
0.10, 42.48 0.12, 25.27 0.15, 14.26 0.18, 19.48 0.31, 14.27
>80
Race
Mexican American
0.05, 42.48
Other Hispanic
0.08, 5.08
Non-Hispanic White Non-Hispanic Black
0.11, 11.00 0.08, 19.48 0.05, 25.27
Other
Gender
Male
1.161 0.869
0.036 0.021
0.05, 42.48 0.05, 14.26
Female
Poverty status In poverty
1.045 1.032 0.980
0.058 0.068 0.028
0.05, 17.02
Near poverty Not in poverty
0.07, 8.03
0.05, 25.27
Smoking history* Yes
1.320 0.942
0.044 0.027
0.12, 42.48 0.05, 25.27
No
*Lifetime history of smoking ≥100 cigarettes.
ing as Other Race, males, individuals living in poverty, and individuals with a lifetime history of smoking ≥100 cigarettes all had the highest BLLs in their respective demo- graphic groupings. In all, approximately 50% of people in the U.S. were found to have a BLL >0.76 μg/dl in 2017–2018. The percentage of individu- als in the >0.76 μg/dl BLL group varied by demographic characteristics such as race, age group, gender, poverty status, and 100-ciga- rette history. The dierences in distribution of BLL were significant for race ( p = .0003), age group ( p < .0001), gender ( p < .0001),
and 100-cigarette history ( p < .0001). These dierences, however, were not significant for poverty status group. These results are dis- played in Table 3. Logistic regression modeling was per- formed for individuals in or near poverty compared with individuals not in poverty (Table 4). Unadjusted logistic regression for the poverty status group with an outcome of BLL >0.76 μg/dl was not significant for either group. With the referent group set as individ- uals not in poverty, the unadjusted odds ratio for individuals near poverty was 1.158 (95% CI [0.903, 1.485], p = .2294) and for indi-
10
Volume 86 • Number 9
Powered by FlippingBook