Without knowing more about your dependent variable, population sizes, etc. this response will be a little generic, and I am no expert in R.

Tests make assumptions about your data and the relationships between the dependent and independent variables. In this case linearity. You can independently evaluate what that non-linearity is, so that you can better understand it and determine what impact it will have on the assumptions of your model. It may lead you to conclude that a higher order polynomial will capture that non-linearity accurately or it may even indicate that regression is not an appropriate model.

Instead of a higher order polynomial you could explore your data using a spline. Piecewise cubic splines are commonly used in finance to fit curves. This [article] (https://www.nature.com/articles/s41409-019-0679-x) provides more information for clinicians.