I would like to point out that topsquark's "Tschrinahusen transform doing more than it should" is not a coincidence. The polynomial in the OP has Galois group $D_4$, which is of order $8$. As
$$|\mathbf{Gal}(K/F)| = [K : F]$$
We know that the field extension over $\Bbb Q$ constructed by adjoining the roots of the polynomial is of degree $8$. But if there were any cubic element in the extension, then by $[K : F] = [K : E][E : F]$ for some $E/F$ generated by the conjugates of the cubic (that'd be atmost of degree $6$) $8$ would be divisible by $3$, a contradiction. Thus there is no cubic radicals appearing in the expressions of the roots of the quartic, hence there is a transformation $p(x) \leadsto p'(x)$ where $p(x)$ is the quartic in OP, going to $p'(x)$, a quadratic in quartic disguise.
In particular the resolvent cubic (I have mentioned what they are in another post) factors in a linear term and a quadratic term.
Just sayin'.