Session: 17-01-01 Research Posters

Paper Number: 69524

Start Time: Thursday, 02:25 PM

69524 - Effect of Inactive Ingredients in Surface Disinfectants and Use of Predictive Modeling on Material Compatibility 

INTRODUCTION

The Centers for Disease Control (CDC) estimates that healthcare acquired infections (HAIs) are responsible for 99,000 deaths in the US annually. Healthcare fomites, such as non-invasive medical devices, are of particular concern due to their ability to transfer dangerous pathogens among patients and healthcare staff.  Disinfection of non-invasive medical devices is of paramount importance, but due to the increasing complexity of medical device design, this is proving challenging. Device manufacturers require substantial flexibility in material properties and are therefore required to utilize a large array of polymers, metals, and ceramics. Incorrect pairing of a medical device and disinfectant can lead to serious material compatibility issues such as cracking, breaking, and even failure. To circumvent this problem, the US Food and Drug Administration (FDA) requires manufacturers of reusable medical devices to provide adequate directions for use where compatible chemicals are listed. This requires manufacturers to partake in time consuming material compatibility testing, which has led to inaccurate material compatibility generalizations with active ingredient classes and has ignored the negative effects that inactive ingredients can have on medical device material properties. Here, we investigated the change in chemical resistance of common polymers when exposed to inactive disinfectant ingredients. We also utilized this data to evaluate predictive models, which can be used to rapidly screen material compatibility of future chemical formulations.

DISCUSSION/RESULTS

We obtained 6 commercial resins utilized in medical devices: 2 polycarbonates (“PC-1” and “PC-2”), PC/ABS, PC/Polyester, ABS, and PMMA. Tensile bars molded from these resins were placed under 0.91% strain, to accelerate material compatibility issues, while 19 different disinfectant formulations were applied. All 19 formulations contained the same active ingredients at the same concentration: >50% alcohol, and <1% quaternary ammonium compounds. Each of the 19 formulations were buffered at varying pH values from 6 to 10 and were blended with different amounts (0-2%) of 4 common inactive ingredients: 2-butoxyethanol, hexoxyethanol, 5-Methyl-4,7-dioxadecane-2-ol, and phenylmethanol. This approach allowed for a controlled environment to assess the material compatibility of  the aforementioned inactive ingredients.  By using a combination of a visual scoring method and tensile strength retention analysis, general material compatibility trends could be identified.

PMMA yielded 0% strength retention for all the formulations tested, whereas “PC-1” showed the highest resistance with 13 of the formulations achieving ≥80% strength retention. For all the materials, phenylmethanol showed relatively high negative correlations with strength retention. Additionally, for the two polycarbonates and PC/ABS, basic pH showed high inverse correlations. Conversely, for PC/ABS, neutral pH showed relatively high positive correlation. The observed results indicate that basic pH and phenylmethanol are significant contributors to material compatibility. Additionally, there appears to a synergistic effect with phenylmethanol and high pH, suggesting that a combination of both may cause material compatibility concerns. For “PC-2”, 5-Methyl-4,7-dioxadecane-2-ol also has a high inverse correlation.

Neural network modeling of strength retention on all 19 formulations and 5 plastics (exclude PMMA) showed a root means squared error (RMSE) of 0.08-0.13 for the training set and 0.08-0.34 for validation set, with overall RMSE of 0.14. Gaussian process showed an RMSE of 0.14 overall. Among the materials, ABS showed the highest RMSE, indicating least predictability. More work needs to be done to reduce overfitting for this material due to its high RMSE for validation set.  In a further analysis, we excluded three formulations when building the model and used those for validation. Neural network showed an RMSE of 0.31 for the three excluded formulations, while Gaussian process showed an RMSE of 0.22.

CONCLUSION

By developing a deeper understanding of material compatibility, medical device manufacturers will be able to provide clear and accurate instructions for use.  In this study, it is obvious that generalizing material compatibility based solely on the active ingredient of disinfectants leads to grossly inaccurate results.  It is our recommendation that inactive ingredients and pH also be considered. Further work is needed to develop accurate predictive modeling, which will be the basis of research going forward.

Presenting Author: Jesiska Tandy Metrex Research, LLC./ KaVo Kerr

Authors:

Jesiska Tandy Metrex Research, LLC./ KaVo Kerr
Alexander Wollenberg Metrex Research, LLC./ KaVo Kerr
Daniela Barrera Metrex Research, LLC./ KaVo Kerr
James Chia Metrex Research, LLC./ KaVo Kerr