Risk quantification and decision
This study focuses on lead exposure-related anemia risk. Therefore, we used hematological indicators to represent BPb-induced anemia. The Hct, Hgb, and RBC were representative indicators of lead-associated occupational anemia and can be used to evaluate the critical BPb levels to protect lead workers [
16]. In addition to common indicators, our study also applied the measurements of MCV, MCH, and MCHC, which were calculated by three original indicators to estimate the possible critical values.
This study found that BPb-related hematological effects of anemia were consistent with previous studies [
16,
22]. Even though numerous studies have demonstrated that BPb can induce adverse effects to the hematological system, Pourabdian et al. [
23] found that indicators of Hgb, MCV, and white blood cell cell did not display concentration-dependence on BPb levels. Some confounders such as menstruation, smoking, and working status may influence the relationships between BPb and hematological indicators.
In this study, the BMD model gave us an opportunity to estimate the critical effect dose for BPb-associated anemia in workers. BMD modeling has been utilized as a more precise mathematical approach for performing toxicological risk assessment [
24]. Compare to the traditional method of using NOAEL (No observed adverse effect level) or LOAEL (Lowest observed adverse effect level), BMD modeling can be further applied to model prediction in the toxicological and epidemiological investigation.
Overall, the BMD models provide more accurate and precise value by using the computational approach, which considers the uncertainty from the epidemiological information and mathematical models. Our results show that BMD
10 and BMDL
10 estimates ranged from 23 to 87 and 15–48 μg/dL for male workers after adjusting for
of age and working years. Our estimation of BMD
10 was similar to those reported by Karita et al. [
16], which ranged from 48 to 73 μg/dL. Further, our BMDL
10 estimates were consistent with their results, which ranged from 33 to 49 μg/dL.
According to the recent statement from American College of Occupational and Environmental Medicine (ACOEM), BPb is still a reliable and wide-used biomonitoring indicator for the health effects associated with lead exposure. The most compelling epidemiology evidence shows that the increased cardiovascular morbidity and mortality in populations with BPb in the low to medium range of 10 to 20 μg/dL [
25,
26]. The most notable effect is the increased cardiovascular risks at these BPb levels, but the adverse pregnancy outcomes and other effects may also occur.
Risk assessment results demonstrated that lead workers are likely to work in unsafe workplaces. Unacceptable hazard occurrence probability was approximately 11.1% for total lead workers. Female workers in the assembling group had a higher risk of 79.5%, resulting in hematological effects, which may exceed the acceptable level. Risk contribution was 31.5% within total lead workers. However, the highest hazard contribution in male lead workers was estimated at only 9.5% in the grinding group. It seems that appropriate engineering control measures and hygiene strategies still play an important role in limiting lead exposure for female lead workers [
27].
To determine whether the estimated critical values of BPb can be used as the novel limit for health management, this study successfully applied a Bayesian analysis approach integrated with MC simulation to estimate the uncertainty and reliability of risk control efficiency. Our approach had also been widely used in exposure analysis and risk decision-making in environmental and occupational health research [
28,
29].
To determine the BPb levels in health management, we examined the simulated BPbs that were set at 25 and 35 μg/dL for male workers. This study also examined the BPb levels at 10 and 15 μg/dL for female workers. Our results found out that there were slight differences when considering these two BPb levels for male workers. Although the ORs and hazard prevention probabilities of a BPb level at 25 μg/dL were higher than those of a BPb level at 35 μg/dL, it seems that the improving efficiency had reached the theoretical level. Therefore, the suitable BPb level of male workers for health management of anemia can be set in this interval. The BPb-hematological BMDs for female workers were lower than male, indicating that female workers are more vulnerable for lead-related anemia than male. We found that BPb level at 15 μg/dL could be used as a suitable health management limited rather than 10 μg/dL.
Our results also revealed that 95% CIs of Bayesian analysis-based ORs all cover the value of 1, indicating that the risk control efficiency may be slightly insignificant. However, the results from hazard prevention probabilities showed that there were nearly 50–70% probabilities of improving the lead-related anemia by restraining BPb limit up to 25–35 μg/dL for male workers. The hazard prevention probabilities for female workers were 85.5–95.0% when BPb limit of 15 μg/dL was adopted. Thus, the anemia risk can also be reduced when a more stringent BPb level is adopted.
Limitations and implications
Our study provides the health protection guideline of BPb level in lead workers, which can be used to quantify anemia risk associated with lead exposure. This study did not focus on the exposure assessment of airborne lead. Compared with environmental monitoring of airborne lead concentration and permissible exposure level, BPb is more reliable and can be used to determine gender-specific health effects of lead. In addition, the airborne lead cannot fully reflect the health risk if the company successfully implements personal protection equipment and industrial hygiene.
This study could not investigate and control the other possible sources of lead exposure outside the work and get the complete information on previous experience of lead-related work. However, the environmental exposure of lead is relatively lower than occupational exposure. This study focuses on the dose-response relationship between BPb concentration and the hematological indicators from regular health examination. The BPb was mainly sourced from the current workplace. Therefore, we assumed that the cumulative BPb that sourced from workplace or other area can be associated with the change of the hematological indicators.
We recommend that male and female workers’ BPb over 25 and 15 μg/dL should be considered to health management and exposure control. Lead-exposed workers need to be continuously controlled and reduce their BPb even if levels are under the current limits of 40/30 μg/dL for male/female workers. Flora et al. [
30] reported that BPb levels of 10–20 μg/dL could precipitate anemia. The possible threshold of lead-associated anemia was also proposed by previous studies, which was approximately 20 μg/dL for lead workers [
11,
16]. By contrast, our study found that BPb-related hematological effects may display gender-specificity due to inter-individual variability. Furthemore, there is no evidence of a safe exposure level for renal and cardiac effects associated with exposure to lead [
31,
32].
According to our correlation analysis, we found that BPb had significantly negative correlations with hematological indicators in male workers. However, this correlation was not observed in female workers. This result might due to the fact that female workers had relatively lower BPb levels compared to males. Moreover, we were unable to collect information about the menstruation or pregnancy from our collated data for female workers. This data gap may influence our risk estimation results. The constructed dose-response relationship was an atypical S-shape in our effect analysis. We observed some abnormal hematological indicators in female workers who had lower BPb concentrations, an effect which may have been due to menstruation or pregnancy. Nonetheless, BPb levels in female workers needs to be carefully revised in order to prevent the lead-related anemia risk.
In addition to hematological effects of anemia, health effects of low-dose lead also include hypertension, cognitive dysfunction, renal, and reproductive effects. Lead may also be genotoxic [
33,
34]. Neuropsychological effects of lead toxicity were also determined to be a common occupational hazard of lead toxicity [
35]. Neurotoxic effects in lead-exposed workers were observed at BPb level below 20 μg/dL, which was a more sensitive toxicity phenotype compared to other lead-induced adverse effects. Ahmad et al. [
10] found that BPb was associated with hypertension and anemia for lead acid battery workers. Therefore, lead workers might suffer from illnesses associated with low-level lead exposure.
Computational approaches are a useful research tool in the health risk assessment of chemical exposure, including heavy metal [
36]. A computational modeling approach was used to estimate airborne lead concentrations by using BPb levels [
37]. OEHHA [
37] used a computational approach and recommended that the BPb level should be between 5 to 10 μg/dL for workers over 40 years of working in a factory. Thus, the 8-h time-weighted-average airborne lead concentrations must not exceed 2.1 μg/m
3. The permissible exposure limit (PEL) is 50 μg/m
3 in Taiwan. Therefore, stringent PEL and personal hygiene strategy should be considered to reduce exposure to airborne lead.
Our study found that female workers had lower lead effect dose than male workers, suggesting that, compared to males, female workers are likely to have a higher risk of lead-induce anemia. In Taiwan, the Occupational Safety and Health Act suggests that pregnant women should stop working in the current lead-exposed workplace and prohibits them to work until one year after childbirth. Kosnett [
8] recommended that women who are or may become pregnant and have a BPb > 5 μg/dL should reduce their exposure to lead.
This study applied the model averaging method to integrate BMD estimates across multiple models. It has been used as an effective method for estimating the model uncertainty in BMD estimation. The model averaging method has also been widely applied to dichotomous dose-response relationships in a variety of risk assessment contexts such as occupational and epidemiological studies [
38,
39]. However, several limitations may affect the accuracy of BMD estimation. Wheeler and Bailer [
38] indicated that the “average-model” method often failed to adequately cover the true BMD when a linear or near linear dose-responses was modeled. The model averaging and the corresponding BMD and BMDL are potentially biased in low-dose scenarios.
The observed average BPb levels were much lower than the current health management limits in this study. However, this study still found some workers with hematological effects even though the BPb were under the permissible level. The proportions of lead working population with abnormal hematological indicators were estimated >10% in each job category. Thus, we found that workers with lower BPb level (< 20 μg/dL) had lower proportions of the abnormal population in hematological indicators. We also found out that some lead workers may still work in an unsafe condition with risk estimates of HI > 1, especially for female workers. It would be better to establish a new occupational strategy and enhance the education and training to reduce the BPb concentrations and subsequent hematotoxicity. Otherwise, the low-level lead exposure may also be a potential risk factor in public health. Our research framework can also be applied to assess other lead induced-adverse health effects in exposed population. Furthermore, a long-term follow-up investigation is urgently needed to characterize lead-associated adverse health effects.