Cincinnati, OH: U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, EPHB 338-05c, 2011 Mar; :1-85
The analysis of the chinchilla impulsive noise exposures evaluated six potential noise exposure hazard indices (HIs) for goodness-of-fit and discrimination. The candidate HIs were the MIL-STD 1474D, A-weighted equivalent 8-hour level (LAeq8hr), Auditory Hazard Assessment Algorithm for Human (AHAAH) in the Unwarned and Warned condition, Pfander C-duration, and Smoorenburg D-duration. The Auditory Research Laboratory at State University of New York at Plattsburgh and the US Army Aeromedical Research Laboratory (Fort Rucker) collected auditory evoked potentials (AEP) from more than 900 chinchilla following exposure to impulsive noise exposures. For each exposure condition, a representative waveform was digitally recorded and archived along with the baseline AEP threshold, temporary threshold shift, permanent threshold shift and histological data from each animal. The exposures investigated the effects of peak level, number of impulses (1, 10, or 100) and temporal spacing of impulses (6, 60 or 600 seconds). The current analysis evaluated the goodness of fit through the use of mixed models that evaluated the immediate threshold shift (TS0) following exposure and the permanent threshold shift (PTS) evaluated approximately 4 weeks following exposure. The threshold shifts were evaluated using six different outcome variables: categorical classification for a 25 dB shift in hearing (permanent and temporary); categorical classification for a 15 dB shift in hearing (permanent and temporary); and as a continuous variable for threshold shift (permanent and temporary). Three explanatory variables were considered with respect to each exposure criterion: the exposure criterion, frequency, and baseline threshold. Goodness-of-Fit: Generally, the statistical analysis demonstrated that LAeq8hr provided the best fit to the threshold shift data for both the permanent and temporary outcomes. The Pfander and Smoorenburg models generally demonstrated the second and third best fits. The Mil-Std 1474D typically had the poorest fit. Goodness-of-fit was judged using the Akaike and Bayesian information criteria. In a separate analysis, the threshold shift data were fit at the individual frequencies against the HIs using a logistic model and the threshold shift as a continuous variable. In these fits, the LAeq8hr was also demonstrated to have the best fit as demonstrated by the Coefficient of Determination, r2. Discrimination: Discrimination was tested by analyzing the Receiver Operator Characteristic (ROC) curves for each HI and the threshold shift outcomes. In this sort of analysis greater area under the ROC curve (AUC) implies a greater ability to predict whether or not hearing loss will occur in the chinchilla. The discrimination results depended on the outcome variable. For the categorical permanent threshold shifts (25 dB and 15 dB) the Unwarned AHAAH provided the best discrimination. No statistically significant difference was observed between the Warned AHAAH and the LAeq8hr, however, both methods were significantly better discrimination than the Smoorenburg, Pfander and MIL-STD 1474D. For the categorical temporary threshold of 25 dB the Unwarned AHAAH, Warned AHAAH, and LAeq8hr indices were better than all the rest, but did not differ significantly from each other. For the categorical outcome of a 15 dB temporary shift, the LAeq8hr index was not significantly different from the Unwarned AHAAH, but better than all the rest. The Unwarned AHAAH was better than three of the rest. Conclusions: The purpose of the interagency agreement between NIOSH and US Army Aeromedical Research Laboratories was to investigate the ability of the several hazard indices to fit the chinchilla data. The LAeq8hr index provided the best fit to the data for all outcome variables, with the Pfander and Smoorenburg indices second and third except in the case of the continuous outcome for permanent threshold shift. In the case of the continuous permanent threshold shift, the LAeq8hr index provided the best fit and the Unwarned AHAAH model had the second best fit. While the Unwarned AHAAH model exhibited better discrimination, the Warned AHAAH model did not exhibit significantly better discrimination than the LAeq8hr index.
Laboratory-animals; Animals; Animal-studies; Noise; Noise-exposure; Noise-induced-hearing-loss; Hearing; Hearing-loss; Hearing-impairment; Hearing-disorders; Impulse-noise;
Author Keywords: Noise exposure; Hearing Loss; Damage Risk Criteria; Impulse Noise; Chinchilla; Equivalent A-weighted Level; LAeq8hr; Auditory Hazard Assessment Algorithm for Human; AHAAH model; MIL-STD l474D