We prove that the coefficient of absolute prudence is greater than k - times coefficient of absolute risk aversion for the utility function if and only if the coefficient of absolute prudence is (3-k) times the coefficient of absolute risk aversion for the inverse utility function. Moreover this is also equivalent to (k-2)-concavity of the first derivative of the inverse utility function.
Amir, R., & Czupryna, M. (2004). On inverse utility and third-order effects in the economics of uncertainty (ECON Discussion Papers 2004/74). https://hdl.handle.net/2078.5/33418