where HI = hazard index, E = exposure level to component i, and AL = acceptable level to component i.
If the HI exceeds 1.0, overexposure is indicated. ACGIH, OSHA and ATSDR’s formulations are mathematically identical. The assumption that underlies the validity of the use of this model is that the shapes of the dose response curves (or more precisely, the exposure-response curves) are similar.
One unique aspect of the Army’s proposed implementation of the HQ is the use of an internal measure of dose (COHb) with an external measure of dose (HCN in air). However, as COHb is a measure of dose rather than response, the combination of an internal and external measure of dose appears to be consistent with EPA guidance.
The Army proposes using the Coburn-Foster-Kane (CFK) equation to calculate carboxyhemoglobin (COHb) levels from CO air data that can be measured in real-time. The real time monitoring of CO provides an exceptional ability to observe the pulsitile spikes (seconds) in the ambient air in the armored vehicle after the cannon is fired. The changes in COHb are much slower, on the order of minutes to a few hours. Factors controlling the rate of change of CO binding to hemoglobin are CO concentration in the air, breathing rate (workload), and the diffusion rate of CO into lung blood. The CFK model has been validated, but as with any model, is not always accurate and typically has not been tested in environments such as in armored vehicles where the air concentration changes dynamically. Due to the fact that COHb changes over a slower time course than air concentrations, use of a running fifteen minute average for air CO would likely be more appropriate although has not been thoroughly analyzed. The committee recommends that the Army assess the validity of the CFK in the context of armored vehicles both using instantaneous measured data and various running averages. This should take into account the typical firing intervals, including the rapid firing sequences and the sequences of infrequent firing that are representative of actual conditions. After firing a round, the gases are at a peak and decline in concentration due to ventilation. When rounds are fired more frequently, exposures increase in parallel. Thus, the exposure is rarely if at all at steady state, and almost always subject to short peaks and declining levels. The ability of the exposure assessment strategy to detect a potential overexposure will therefore depend in large part on the appropriate selection of an averaging interval. The use of running averages for HCN exposure assessment copes with this difficulty by ensuring that a short peak exposure is not missed, i.e., selection of start and stop times for averaging are essentially moot since every configuration is calculated. The use of the internal measure of dose, via the CFK equation’s calculation of % COHb likewise copes with the highly intermittent exposure by calculating an integrated measure of dose through the biomarker, wherein the biological process serves as the means of integration.
An alternate to mathematical models such as the HQ is the use of physiologically based pharmacokinetic (PBPK) modeling. PBPK has been used to understand interactions between individual chemicals found in a chemical mixture once the chemical mixture has entered the body of laboratory animals or humans. Inhaled solvents have received the greatest attention with PBPK modeling. Solvent toxicity of a chemical mixture such as central nervous system effects may be governed by the brain: blood partition coefficient for each chemical, while other toxicities are mediated by the formation of reactive metabolites. The use of PBPK models to predict the metabolic clearance of solvent mixtures from the body has received the greatest attention (Haddad et al. 2001) by quantitatively describing the competitive metabolism of each solvent by the same enzymatic system (e.g., P450 isoforms). The impact of solvent mixtures on individual solvent pharmacokinetics is governed by the exposure level of each solvent in the solvent mixture, chemical specific properties of each solvent such as its affinity for the metabolizing enzyme and thermodynamic properties (blood: air and tissue: blood partition coefficients).
While a PBPK model has addressed CO as a byproduct of solvent metabolism, currently there are no PBPK models for CO and HCN mixtures. In the case of CO and HCN, a computational research effort designed to understand possible mechanistic interactions (hypothesis generation) between CO and HCN is warranted.