The study conducted by Li et al. (2013) demonstrates how challenging studies in occupational work groups can be to conduct and to interpret. These investigators compared the performance on a battery of neurological tests of vanadium slag processing workers in Panzhihua, China to that of workers in a cold steel mill plant located 4 miles away. Differences in performance of the two groups on a number of the administered tests were noted and attributed to differences in exposure to vanadium compounds, but other important potential causative factors also need to be considered. The relatively large group sizes were stated to be adjusted for age, smoking and exposure to noise; however, the two facilities have other important differences. Vanadium slag processing involves a multi-step process including grinding, roasting, leaching, ammonium precipitation and smelting (Taylor et al., 2006, Sun, 2012), which results in emission of heat and a number of substances including particulates, carbon monoxide, ammonia, sodium hydroxide, sulfides and manganese, into the work room (W.X. Jiang, personal communication, November 14, 2013). During a recent visit to both facilities by an author (JAM), the operations in the cold rolling steel mill were quite different, without apparent dust and heat and are an unlikely source of these substances.
The study did not measure exposure to either manganese or lead, and it is stated, “…our observation may be limited by the interference due to co-exposure to these two metals commonly seen in the workplace.” In particular, manganese, a well-known neurotoxic substance (ACGIH, 2013, WHO, 2008) is present in high concentrations in the slag (approx. 10% MnO vs. 15% V2O5) (Wang et al., 1998, Wang, 2004). Manganese has been shown to have neurobehavioral and neurophysiological effects at relatively low levels in numerous worker studies and occupational exposure limits for manganese have been based on subtle functional effects (European Commission, 2011). Thus, without a reliable adjustment for concomitant exposure to manganese, it is not possible to attribute the small effects observed by Li et al. to vanadium or to any other substance at the study site.
Vanadium atmospheric levels in the slag processing plant are reported by Li et al. to be a time-weighted average (TWA) of 0.216 mg/m3 and described as “low.” It is not reported what period of time is included in the TWA, the particle size sampled, the number of individual measurements or the location of the sampling. With such a multi-step process as that in the facility studied, individual worker exposures to the various chemicals in the workroom would be expected to differ significantly among subgroups of the cohort tested. Levels of 0.1 mg/m3 vanadium and above have been reported to cause irritation of the respiratory tract, including nasal and pulmonary symptoms in workers, and these effects are the basis of the derivation of occupational exposure limits to vanadium compounds by the ACGIH (2009). Therefore, levels of 0.216 mg/m3 vanadium as reported by Li et al. (2013) are quite high and prolonged and repeated exposures to vanadium and other irritants may have contributed to the subtle performance differences on the neurobehavioral tests reported by Li et al.
Testing of workers for evidence of neurobehavioral effects can be complicated, and the test setting and experimental design can have significant impacts on the study outcome (Anger et al., 2001, Williamson, 2007). In the Li et al. study, the workers from the two plants were tested at different times separated by a few months (Q. Zhou, personal communication, November 13, 2013). To control for differences associated with the time of testing, subjects should be randomly allocated to the testing schedules and tested blindly to remove the confounding effects of such factors (CEC, 1997, Lavelle et al., 2012). Other factors between the two groups of workers, such as general working conditions, physical activity, education, mental or physical health status should also be considered as they could lead to differences in responses to the tests. The authors state that they have eliminated from the analysis test scores that were not thought to be physiologically plausible. No further details are provided about the number of eliminations, whether this was done blinded to the exposure status, or about the potential impact of the eliminations on the findings.
In the Li et al. publication, causation has been attributed to a single factor, vanadium. However, other significant risk factors should have been considered and the relevant available dataset utilized in the analysis. Clearly, exposure to manganese is not excluded as causal factor in this study. Analysis of the dataset should be made using statistical methods to assess the relationship of the test performance to exposures (CEC, 1997). This facility does maintain records of job classification, job history, and it measures exposure to particulate dust, carbon monoxide, sodium hydroxide and sulfides in addition to vanadium (W.X. Jiang, personal communication, November 14, 2013). Therefore, these factors could be considered in future analyses—but inclusion of information about manganese exposure is essential.
In conclusion, the data presented by Li et al. are interesting but insufficient to identify the cause of the reported differences in neurobehavioral scores. Further analyses and measurements are needed. It is encouraging to note that the investigators and the company have an interest in ensuring the safety of the workers and continued research in this area is important. The data set provides a unique opportunity for further study.