For every quadratic factor x2+bx+c, we have Ax+Bx2+bx+c.
For every repeated quadratic factor (x2+bx+c)n, we have A1x+B1x2+bx+c+A2x+B2(x2+bx+c)2+…+Anx+Bn(x2+bx+c)n.
Example: We have the decomposition
2x3+5x−1(x+1)3(x2+1)2=Ax+1+B(x+1)2+C(x+1)3+Dx+Ex2+1+Fx+G(x2+1)2.
Finding the coefficients A, B, etc. is also harder. After cross-multiplying, we can't just plug in the roots of Q(x) to get the coefficients one at a time, since the quadratic factors don't have real roots. We either have to plug in lots of different values of x, or compare the corresponding coefficients of 1, x, x2, etc. Either way, we get a system of linear equations to solve.
Finally, solving the integrals at the end is harder than with linear factors.
To integrate ∫x(x2+a2)ndx,
we can substitute u=x2+a2.
To integrate ∫dx(x2+a2)n,
on the other hand, we can substitute x=atan(θ).
These complications, and how to overcome them, are explained in the following video.