M. Nunnally, 3M Co., Cordova, IL.
Utilization of Bayesian Decision Analysis (BDA) is an extremely useful tool when validating exposure assessments with air sampling data. Using the AIHA Exposure Assessment and Management Model, the Bayesian’s probabilistic outputs drive exposure control banding with synergy. The probability of a “decision statistic” falling into one of the exposure control bands is clearly communicated graphically using the exposure control bands recommended in the AIHA Exposure Assessment and Management Strategy. When assessing multiple process tasks with accompanying task based sampling data and then communicating exposure risks to various levels in a facility, BDA is a powerful tool for the plant industrial hygienist. The presentation will cover basic applications of BDA for use by industrial hygienists in a manufacturing environment.
R. Bowen, University of North Carolina, Chapel Hill, NC; M. Harper, NIOSH, Morgantown, WV; J. Brooks, University of Alabama at Birmingham, Birmingham, AL.
The use of deterministic models is gaining in popularity to estimate occupational exposures to airborne chemicals in the workplace. The UK Health and Safety Executive developed a risk management tool that estimates toxicological hazards and includes a deterministic chemical exposure assessment model. This tool is called COSHH Essentials. This paper presents a case study evaluation of the COSHH Essentials deterministic exposure assessment model. Retrospective personal measurements of methylene chloride (dichloromethane; CAS 75-09-2) and prospective personal measurements of isopropanol (propan-2-ol; CAS 67-63-0) and acetone (propan-2-one; CAS 67-64-1) were compared with the COSHH Essentials predicted exposure ranges. The model’s ability to adequately protect worker health and to adequately assign appropriate engineering controls was evaluated. The model adequately described employee exposures for methylene chloride and isopropanol but fell short in describing the acetone exposures. The estimated probability of an employee’s average exposure exceeding the upper limit of the COSHH Essentials predicted range was less than 0.05 for the methylene chloride and isopropanol exposures, but the estimated probability of exceeding the upper limit was greater than 0.05 for the acetone exposures. For acetone, the model suggested the implementation of local exhaust ventilation, and for isopropanol the model suggested the implementation of containment, in both cases with the aim of further reducing exposures, even though exposures were well controlled below applicable standards by general dilution ventilation alone.
M. Weeks, C. Simmons, F. Boelter, Boelter & Yates, Inc., Park Ridge, IL.
A series of air change rate studies were performed in single-zone spaces using both carbon dioxide and sulfur hexafluoride as tracer gases. Testing methods were based on ASTM Method E741-00, Standard Test Method for Determining Air Change in a Single Zone by Means of a Tracer Gas Dilution. Sulfur hexafluoride is often called the “gold standard” for performing air change rate studies because it is easily detectable, nontoxic, and inert. However, the technology available for reliably detecting sulfur hexafluoride, as well as the gas itself, is expensive when compared with the cost of other common tracer gases. Using carbon dioxide as a tracer gas poses the potential problems of high ambient concentrations and the presence of additional generation sources. Air change rates were calculated for eight single-zone spaces using both gases. Results show that in situations where there was an air change rate fewer than 10 air changes per hour and no additional or unaccounted sources of carbon dioxide were present, air change rate results using carbon dioxide and sulfur hexafluoride were consistently within 1 or 2 air changes of each other. This indicates that although using carbon dioxide as a tracer gas can be difficult because of the presence of confounding factors, in many cases it can appropriately be used to determine air change rate in a single zone. Higher variation occurred with ventilation rates greater than 10 air changes per hour. The variation in results and possible causes of the variation will be discussed, along with the practicalities of applying the method to the two-zone model in work areas with low air change rates. The conclusion is that for single-zone work areas with low air change rates and no confounding factors, the less expensive carbon dioxide tracer gas correlates well with the sulfur hexafluoride gold standard.
S. Parsons, CSIR, Pretoria, South Africa; P. Jensen, C. Wells, CDC, Atlanta, GA; M. First, E. Nardell, Harvard University, Boston, MA; K. Weyer, L. Roberts, MRC, Pretoria, South Africa; E. Mathews, North-West University, Potchefstroom, South Africa.
Tuberculosis (TB), one of the world’s greatest killers, is predominantly spread by the airborne route. Multidrug-resistant TB (MDR-TB) has emerged as a global public health threat despite effective drugs and disease control strategies. Little is known about MDR-TB transmission and the efficacy of necessary interventions for infection control, particularly in light of the global HIV/AIDS epidemic. The state-of-the-art Airborne Infection Research (AIR) Facility, recently established in South Africa as a joint partnership between the MRC, CSIR, CDC, and Harvard University. The design of this unique facility will facilitate scientific studies to answer fundamental questions about the infectiousness of MDR-TB, the role of HIV, and the effectiveness of environmental controls to curtail transmission, and will provide the scientific blue-prints for design of safer health care facilities and the development of improved building and construction standards. The facility involves extraction of infectious air from patient wards to exposure chambers housing guinea pigs that serve as quantitative samplers of human-generated infectious aerosols. Measuring the number of guinea pigs infected over time, linking guinea pig infections to individual patients by means of molecular techniques, is our unique sampling strategy. The apparatus supports the many experiments on the effects of varying ventilation rates and environmental conditions such as temperature and humidity on transmission dynamics of aerosolized infectious particles. Engineering interventions against the transmission of infection, such as ultraviolet germicidal irradiation,, and other electro/mechanical interventions will be tested and evaluated. The electronic control system is entirely web-based at controller level running on an open SQL database for reliability and accuracy, providing international research collaborators access via a Web interface. The state-of-the-art facility and its environmental control installation required for patient and health care worker safety and animal care, as well as to satisfy essential ethical protocols of these studies, is presented.
T. Bowie, M. Keinath, E. Miesner, C. Stubbs, D. Daugherty, ENVIRON International Corporation, Emeryville, CA.
Johnson and Ettinger (J&E) introduced a screening-level model for estimating the transport of contaminant vapors emanating from either subsurface soils or groundwater into indoor spaces located directly above the source of contamination. Typically, the J&E model has been applied to vapor intrusion into homes using default input parameter ranges developed for residential scenarios in different climates. These parameters, however, may not be appropriate for large commercial buildings. We present a validation case study of the J&E model by examining the impact of input parameters on predicted indoor air concentrations at several large commercial buildings located in the vicinity of a large chlorinated solvent groundwater plume in Mountain View, California. The U.S. Environmental Protection Agency’s (USEPA) User’s Guide for Subsurface Vapor Intrusion into Buildings (2004) indicates that the scarcity of empirical data available for either bench- or field-scale verification of the J&E model’s accuracy requires that the user consider the variation in input parameters and explore the impacts of assumptions on the uncertainty of model results. Therefore, empirical parameters were validated by comparing indoor air concentration predictions with indoor air samples collected and analyzed in accordance with USEPA method TO-14 SIM. In particular, we examined the applicability of default J&E input parameters to large commercial buildings by comparing modeled and measured attenuation factors. Whereas some of the default input parameters were reasonably accurate for characterizing vapor intrusion into commercial buildings, many required modifications to account for the differences in commercial settings.
P. Hewett, Exposure Assessment Solutions, Inc., Morgantown, WV; G. Ganser, West Virginia University, Morgantown, WV.
Exposure datasets often occur where one or more measurements are below the limit of detection (LOD). The purpose of this study was to test various standard censored data analysis methods when applied to low (i.e., <20% censored) and medium (20-50%) censored datasets, and to compare these results to those for beta-substitution, a newly developed substitution technique. Using computer simulation, the bias and root mean square error (rMSE) for the two commonly used substitution methods (i.e., LOD/2 and LOD/sqrt(2)), beta-substitution, log-probit regression (LPR), and maximum likelihood estimation (MLE) were determined for the scenario where there is a single LOD. The parameters estimated in the computer simulation were the distribution parameters (i.e., geometric mean and geometric standard deviation), the 95th percentile, and the mean. The MLE method has substantial bias for small n, but for the larger sample sizes is nearly unbiased. The bias for the common substitution methods can be positive or negative, depending on the sample size, fraction censored, and geometric standard deviation. LPR is substantially biased for small sample sizes. The bias and rMSE for the beta-substitution method is nearly identical to that of the MLE method for all combinations of sample size, fraction censored, and GSD considered. The sample estimates for the parameters selected tend to be substantially biased for n<10, so it is important to understand the direction and magnitude of the potential bias for each method. The common substitution methods are biased for all sample sizes and should be used with caution. LPR could be used whenever n>=20, but is clearly not suitable for small n unless only the mean is of interest. MLE is preferred above all others, but the beta-substitution method, which is easier to calculate, is nearly equal to the MLE method in terms of both bias and rMSE.
Posted May 30, 2006