Quantitative headform fit evaluation and predictive modeling to assist with selecting N95 filtering facepiece respirators to mitigate respiratory hazards.

Ensuring that respiratory protection is effective for all can be difficult if limited resources are available to assist with selecting a well-fitting respirator model and user guidance. To better understand how various N95 filtering facepiece respirator models fit on a variety of face sizes, a quantitative fit evaluation was performed on 12 different N95 respirators distributed by the Strategic National Stockpile using five manikin headform sizes representative of most of the U.S. worker population (540 total tests). Manikin fit factor results varied depending on the respirator model and headform combination. Four respirator models achieved passing fit results across all headform sizes. Predictive modeling was then initiated, where the headform most closely aligned to an individual's facial dimensions is determined and then used to identify N95 respirators that may provide an acceptable fit. A multinomial logistic regression model was trained and tested using NIOSH's 2003 Anthropometric U.S. Survey and was found to have an accuracy of 85%. To address potential risks associated with predicting only a single headform size, a modified model allowing for multiple headform size predictions was also assessed and found to have an improved accuracy rate of 98%. With further human subject validation and field testing, this modeling approach could be used as a tool to aid in making the fit testing process more efficient, less burdensome, and better enable individuals to use respirators that fit more effectively, thereby adequately protecting them from hazards.