Do we underestimate risk of cardiovascular mortality due to lead exposure?

Abstract

Background: Studies using data from the National Health and Nutrition Examination Survey-III (NHANES-III) have demonstrated significant prospective associations between blood lead levels and increased mortality. Bone lead represents cumulative lead burden and thus is a better biomarker for assessing chronic impacts, but its in vivo assessment requires special K-x-ray fluorescence (KXRF) instrumentation. Our team recently developed an algorithm predicting bone lead levels from a combination of blood lead levels, age and other socioeconomic and behavioral variables. We examined the associations of our algorithm-estimated bone lead levels and mortality in NHANES-III.

Methods: We included 11,628 adults followed up to December 31, 2019. Estimated tibia lead and patella lead levels were calculated using our prediction algorithms. We used survey-weighted Cox proportional hazards models to compute hazard ratios (HRs) and 95 % confidence intervals (CIs).

Results: During the median follow-up of 26.8 years, 4900 participants died (mortality rate = 1398 per 100,000 adults/year). Geometric means (95 % CIs) of blood lead, predicted tibia lead, and predicted patella lead were 2.69 μg/dL (2.54, 2.84), 6.73 μg/g (6.22, 7.25), and 16.3 μg/g (15.9, 16.8), respectively. The associations for all-cause mortality were similar between blood lead and bone lead. However, the associations for cardiovascular mortality were much greater with predicted bone lead markers compared to blood lead: for comparing participants at the 90th vs. 10th percentiles of exposure, HR = 3.32 (95 % CI: 1.93-5.73) for tibia lead, 2.42 (1.56-3.76) for patella lead, 1.63 (1.25-2.14) for blood lead. The population attributable fractions for cardiovascular disease mortality if everyone's lead concentrations were declined to the 10th percentiles were 45.8 % (95 % CI: 28.1-59.4) for tibia lead, 33.1 % (18.1-45.8) for patella lead, and 22.8 % (10.4-33.8) for blood lead.

Conclusions: These findings suggest that risk assessment for cardiovascular mortality based on blood lead levels may underestimate the true mortality risk of lead exposure.

Figure 1.

Hazard ratios (95% confidence intervals) of mortality from all-causes and cardiovascular disease (CVD) comparing the 90^th vs the 10^th percentiles of lead biomarkers by sex and race/ethnicity. See Table 2 for more details of modeling. NHW, non-Hispanic White; NHB, non-Hispanic Black; MA, Mexican American.