A New Method to Adjust for Ordinal Variables With a Pre-Diabetes Case Study